73".. a: "I" mum'- ‘ urfIII 11311113111111? «3,. fifimfihknr ".3, 3II3I3I}I' .3;3'3 3III."I.III ‘II‘I3IE 353113119410I'3I33H3EIIIF. 1‘11“” 331131 231313: :13I3231211I- I. 3' I I" 1113133333,; 11II1'1111‘11'111111111111. 33 3 Id” “IIwwwwwhmemflwmflI 1V1‘fiquIIIh ‘ . 1' fiWIXH fiWIflan'1flflgfififlfifif; 1‘1 333333 I 3 33v I313”; 0'133‘3 I' I I I! ”3.33.”...3‘333 , . (33 "33 :3 33 11%|“ 1 |.‘.:.\M n ‘11? 1] 1 #111,. 1‘11“ ‘3? 1.1g .1 I1 11131 |:Il' I 3"1N:1313q11131': 1I11{:I'I1. 11111111 111'11“11‘“L 31111‘1111311‘3”4‘1“”411‘11111 . . . II:- -..:-.- {:1 {11111311113331 1.111 I-1‘,I‘ I, 31333}: 1,131'III.III.333 “33333333 Ill-I11 .11 II11.13|L,I!/y .3 “1'11igpé'3! ,I3‘_33 13131311141313'11‘111I'11‘31 ‘ 333?.1'I'5131111, 7 .1I1,E:‘113;3I~_333 ". A 111:1;1U1" 13:33:31.! 133; I, 11111!‘ ‘.3‘, gig {111v IH" II" '. 1.1:}? 11114111) I“! 11t::.,313;§ 3 IIIIIIII, .31 1'}I}|1"‘l[11‘1f‘1111.;."; .I1I'I " 311311113 ”[1511; " 1‘1‘1I'IIHIII I1-.I13'IIH‘1 1131‘; 1'1- ~" '11 g 1335;1'?1'1f' 15111., 11b. 1: 7.34132} Ilka?! “gm- . Iflymtg.§~'.¢~ga11I.I-%mmIII2wImII HiqifirI. - $1.; .I,"-I"I"-.1!II',-L ..I.. .. ' 1111MJ3EI1111' 31% "11131311111111“ I‘E1I' 1» A“). ' :? 3331“" Qua“ “11¢; I‘IIIII‘IIIII {111' 22516 . I 5.. 1333 .~T'.LI I:\\,7cI ,3,':‘: ‘- 37 a‘7‘t‘Vf‘pt; I 3331 .[ 1‘1" 3 [3&3'3193: “h“.‘t‘g‘ ‘1!“ (zutrk 1.8 .$343.33.- 1%?th .‘féu'i‘h' W: ”41. D ”1133;: “"" I ' I - ' I . .,- . .,. 5-: 'fi 1‘11} 11.4%.»! :19 run-":1 I‘m» I1"':5I IE5” “WV“ I3I“ ‘ .II, I91“ ~£IEIIIIIIIII11 4:11;: IIII‘I'II '1" “‘1' II - 1‘11“" 1’III‘IIIIIIUII‘II ' f 11f x ’3‘ , 3 I': ;:!'1"$rlir3¥ {H A" k 11 “HA“?! ' '3 ”115;. .III‘Iz1-I‘r'1 1'" ,1.“ WWII-1'" 11".” “‘ "113111111-.01'.‘I"1‘11'I"1':'§"-55Iu W 1-‘111 ‘ I. .I ' II‘1‘I7'“‘I" 111113397111111111'11111 .5333 4:33,: .1433 ~,{“""11111;I;3‘11‘;1I 7I14ggi;.3:33.313;;3:333,33333v 3333;33:3[I1333I' mug? 1| 54.3..." . 335,; 3_33 1‘ I331”! 1.3 ‘ ‘1 I I I . ‘ . .II I" ' I' .‘.‘- . I»; 1, IIIikIII-c-I'. pm; 11551.,4 .1 :_, I , : 33333113 3333'3-II1I3111333..3113:3 3:13:31 13111753 E1213 .3, 1I3I3‘I§I"I1 1131,13 II ‘ "11:11'3 3 311'1’1Illi' '1'.- 'r' 1‘":- “'13:; I. 13.1“ 1 ' “'I '111 1 'I I ,'; 3 3 ' . I" g. ”[3,, _ 973.131. "ir|1‘117‘r'é ‘1‘! '2g:;'.1111‘-1’|II1I11V'113 ,‘3‘, "I'1-E'3 11:131.}? 3'7 C 3’33 1‘3"} 31.3 34- 333: . 31.31.33 3-}: .‘ 33. 1'11};- ',‘3J1 \‘i3, 1,131. 11} 1:533:33 1». 3:713“ ,r 3“; “1.! 33‘:31_3.3 3, 3 3 3 3 ”‘33.:‘333 213111.11“; 111'I.( LIEL;"P:L.1V|3.111E‘I,‘l."‘:1‘1:E;§Eg’!li!;jdil1lwrhf 17111111111;:1’111;11;T;?r‘“11-h...“ 11,333.33 3:, 1i, ! 1.1234131. .‘3 1.3.3- '3I:‘I I' 11.11%!” 333333131 3. t3333-33 113111If‘nn 33 “3,73, 13I3‘3I!.II- 11 D .r ’7J1333 3‘13." '43:, 3.111; 3-3 13..33 . 3 ,1 7 ' -II , 3 I. -, I I .3; {é ‘2 3 (32:3,, 5' ‘3: {33313133 .3 I. 3 333 33 ,3 3 33 II.IIII..I-II3I3333‘333 3t 3,0.‘4'13‘ 31434311 3;‘3LI’3I.I 3331‘ III'LI,I13-.I133.3I3:I 3. I3 , 2:8)9111131'71111'1111k1:‘1113‘11u.“ ”"11“1211111H'1 3 11'1',111"'1I;3": 3333:1213 :33 “#133334 $313313 3.333 [[3 ?L '11"? l1! ['3 13 3 3343 3 1 , . 3 3 1.3,: 354 ' I II If! W1: ‘:‘3'3\111'I11i1'1‘1111klr '1," 1.3.31,“ [-11 -.I I”z‘-1'3'-I ,.- .IL;I, ' . . ._ 1.: I 1311141'411 33:33., { 3311;313II 'I1 "1 I'fIIII ,1'MIII.‘I‘. .|3| -;i I II I. I . I 311L119 $3132.; 33: 33333 393333;.I_L'3'VTI,13:_.§‘II’.’;“3M3... £13 11.1111131111111‘” , 5.3335333 . .. II. ‘I’. . WM I"V‘ .I. . ; . . I~"-‘I.‘- 1I'ILIL I .',‘I . I: . ’ , " I .1 .11111;?“1111'331311311111"2.613.191. IIIIIII'1II:’,‘1 1‘ .‘11I”;'I'III 1» 1"I E “13111’1I’Lv III-‘1 ~11" 1‘“ $11,111,13Lifh' ‘ ' I ‘. '1 ". I1I'I 'I-IL' . 1 I.“ . II .I. (.1 .Z-IIII E13‘Il“"-? I, I1: II ' 1 L I ' ‘. : 3131351.? $111131-11111;1;21 ”1113113311333313113333313.1113 1.11:”:“1JI1113.1131313‘3313'3'31. '-3:.‘3‘11;1’1l1111g.:)5l1 :3 ”“333 1 3 3 l3 :3 i I 1‘: 15.313 1 1:31.3111‘“ 13133.1}: 1111 W'::‘ '"3 fl. ' ‘11:?“11333 3111‘ s, 1. 3 ”'1"; . 3 1" 7| £3? ”“1"13373‘ o o 31. 'j‘SI‘ '11:" ‘3 I. g :1;' . . '3 .‘7’31 11., I 1111,111111131 1:;111!111.o" E1 "11110113611111 I '1'111‘1111‘11011 V '1111.111 11 I "fill 1.13311 1:1'111:\I11:1.1HI ‘ E1111 ":11 I'M,"- I .1" “117.11". I1} ”131°11'11" ' 1-',I - 1 '4 1 4 I H1333” 1'1' , - I . - ' ”V 11".] t“.1'.:u l'311, y'17I311111‘13“. 'I'11.|': I I3I133r3;1 7111111111)?“ ‘1... 1:1 . '1‘ 1;?» 1 I1I1II1' .I 3313-1 IQ, 11121111". 3 , ’I' ’I,I,I;~II'I.I'-. I, 353w 1,1.» ...,1I '11111'3‘111'1 1 ML! .. 33:8-13 11.3 ~II'n‘I I 3.1' 33.3. 3. , - 3 :333‘331333133333'1I133'.311333113:|J313b3313 3.3I3I..3I3-I3j,3_j3'3. ,31',1-. .3433 3| 33:3I:I3I3 ,3I‘33333I33 3 I3I .. 3I3II333.3 ,3’3HII333I3LI333I3 33 3'33 333 33 33 33 3:33.333? 33 3:3 ,3. . , 3 33 . ' 3'II‘ 3, '3' -' I,'._3I‘.33I,I 33 . 3'1, II Win. 53' 1.,‘33.I3233I 33333 , II_.I«. .I . ' 3‘3PI1 1'11. ‘ 4"I 1‘ 111 'I“ :11 '1‘1' ‘1‘ ‘1 '151 173;}; II3'133‘31‘}I3‘ 31”} 333 ‘.‘ .3311I,33I,,1I;3..31.3‘1 .11", 1-33 IA 1 ' 1 . .I. :3}! 3.3 ‘I3 , 'V $3. 3113." ',. I . . I333I.v I 333 3 3 '8'. I ‘0‘» 3._ 3 '115'11‘ifi1111111:11:1(11'13‘311'1'“ 11 '1 111‘ ‘314‘11“ ‘1: l1 1 ' 1A1'1":“11'.. My". ."II ‘I f I (311v ' ‘ , 171-111-II‘1;1 ..1 31-" . 1‘ "I I ‘2 I e, , ’ I': E $31” .,'I 'I" " x‘I Ifzgzfiuw‘vgkl 5;}: t3 3‘13‘II'I‘1 1III‘I;33 337339131 I;3'3:1:II1T 3 3333.133333’3133333L3 :, 3 33‘3 . 3. ,3 l3 Iy."‘.7‘1"1 '1‘P1i‘1313 111'11 3 "Ii-13.; L; NIL-I] “"73”??? '1111' ,. 3. ‘ '. .I 111:1'11'1' '1“.'III';I' ‘I: . . III?" I. III» II. I. ,I3 I - I ., I» II . .- I I II. , I ‘ 5’." ’3“ 43.3” 13 I -.§r1.331333 III I3 I 3.3.33 ”9,333 3313i‘III’fI L 17 IN “I , 333, {3114 g1111 3 3 ~ 33 I, I I3‘ 1 3333333331 33333233 1,.;111"’151I1"13";1111'13‘11'11'11. [33313331133433 3. 31, . 3 :1 331‘ 33.33 p I. ;.'3.3 f -3 ~12 .. 13‘. I3 3 .3 .3 3.. ' . _" .I a I '13 v I." '1 A h [ 1’ fl. 1 ' $ , J II '3‘,“ . 3 ”.13 ":111'1‘1111'1‘ 11.11 133? , 3; 1713:3I'1 '33.. ,3",I'}I‘31j‘3 .3331,.3 114,111.311:13.11'331'31’1111411.1+“ 11‘31’7r I 131,3... ;3( 3 1413'. 3.33 3: 3:: 3333.333 . ,, ,1 3_I_S' :3-33,33:3 3,: .' 3. ‘ 3 v .3 , 3 ,3: I V3 .333 I ,:.3..3 .I I-IIII- 'I “I -: ‘I.1'1:1‘II11I'I III; I' 1 I .II- "#1“! I» 3-3.. I I . ‘. . . . I-: . an. ., - 1 13":3'1;11111 1W" .. ~.I-.I I I II 1331, 11111 "‘3;[ I .33-1'113I'I‘11"'f""r1'13w1 . . 3 t I b '17 l ,‘3111.11|1.! . - I.. , 1| IMI I! i . I‘A' '1'111' 3.3 “131333 III 3 . - 11. r. 1 [‘5'1'1I'31‘737‘fill3‘ ,3.-‘..'II{',I3III3II ' 2‘ I51I1’3‘31II‘I'E1II'I$1111.11!II"I.‘I"‘.f?~-I"-II.’II I:- ‘III- MW "1" I ‘I I I” ”'11 .,,3,I3I,;3 '33111.1333£‘1III1’II‘ aim, i=v§13:3,3.3‘3: II; . =Im»wmnIb~”vanNnII .I ‘IIWI .. III ‘: ’ I A uh» . . .,-I .1-II1..'III:1I .4. 1. - 4.; 51131.! ”11111‘11 I.‘1.I.~’ 1.: 1n'131z'111r11'l’13131'1'r , ‘I IIIIIIIII . . "1"15‘113'1' II'II'1 IIII IEI .._ .I h... .I I ., 1 1.1"» ,;£ .1. 1113111333121!!!133111.11141115131331311::IEEEFI 31 "1 '11:; 311'»: MI“, 1',.11’.;5IH:'§11 E "IIIJ'T'W." "MI MI I ’1 .II ' 1' r ' 1| 111 I I111 1‘ I.,I1 ,1 I13“ Thu' I I ,I3 3'd31. I .3“ I II. II‘ {1 pl" . ,t3 .. 3M3, 3. I I III.IIIIIIIIgIII.II-I=I.IIIIIIIIIIIIIIIIIIII... III. III... III-u. ,, F :3 m 3 .3 .I,:33t, 11161111111 ‘3..3I "'3'33311‘51'nm ’ ’1111'1111 ;‘I ,,' 11"1'111" '1‘3113'1131'1115.1111311... "5’1 1137‘ I33“ ‘11131‘1 1111‘" 1111-1111”4 "' 1311131511 ’1 111‘ '11'11! ' 1'th '1111‘IIII1IJI' $511‘.EV‘ . I'I.'II1I‘II'I. I17 ‘I"'I'-i ’ ' 1.7341“. .3 -. I 011331.“ 3.5.” .2 233‘. :1 11181111111111. 13 33 3 3911911, 1,,1)“; , "33.31-31.13“; 11:11'1‘22'1‘1'3-1131 u .535 I ' .111; IjI; 333333 I ‘.~ 'g.- ,. 3.32.3», ‘. 2H Ifi‘fil‘vflh‘ $‘~. I 1B If' 1;: ’I II- . HY}. I I-. 113; .II' 1 -L‘;' »'I " . I. I M ,- -, 11:7! , 3. an} 'gn'flg‘ 1. " ' 1 '; 0'74 ,Itgh‘ 41!! "I 111:5'71' '1l‘7‘ ' "1 1 H" " 1‘ L .11? ”#1, ‘I I.3, ' h 1r1l I, - v .1. , .1111’ ' ',o{ -1 ‘25:]: IfiMWWIWMWwfiwwwIIIMMWIMIWW4W4wwWWWWVWflw . 5" .fimn fiinWhI‘ImnhiWhu II I uh‘ ' ‘1 wwI " ‘JVJW?LT IMWWWWWMWW%flHHIMW“WN§wI IIII W’m.~W%% .".I praufi‘. I" ' "III .. ' .I . III I" I :Igv i 111I111.IIIIIII1IIIII1IIMWII “I”1II 'II‘1 11IIIII II I91W1fi~1 III III 11 '41:.1111I10 1” 7.11.1.1l1'1112‘.111."u!§ 11' 111 761:1 (1111111 H1 1.1-“ ”.1 . 1.7153111.“ )NBlSHCUOI ”W" ‘ l\\\\\\\\\\iji\”\i\ \“\‘\\\\\1.\\\.\ low W Michigan State 31293 L University This is to certify that the dissertation entitled Explaining and Predicting Supreme Court Decision Making: The Establishment Clause Cases, 1970-1986 presented by Joseph Anthony Ignagni has been accepted towards fulfillment of the requirements for Ph.D. degreein Political Science / if ‘ {/(z / Ma' essor Dang/”2’73 flfé M S U it an Affirmative Action/Equal Opportunity Institution 0- 12771 PLACE IN RETURN BOX to romovo this chockout from your rocord. TO AVOID FINES rotum on or baton doto duo. DATE DUE DATE DUE DATE DUE . I 'U 5 sea}! A 'i rsEP 9 6 i373? ff? 0 Q 1°"5 ER 2 7 1% I | MSU Is An Atfirmdivo AdiorVEquol Opportunity Institmlon EXPLAINING AND PREDICTING SUPREME COURT DECISION MAKING:THE ESTABLISHMENT CLAUSE CASES, 1970-1986 BY Joseph Anthony Ignagni A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1990 Coafi934 ABSTRACT EXPLAINING AND PREDICTING SUPREME COURT DECISION MAKING: THE ESTABLISHMENT CLAUSE CASES, 1970-1986 BY Joseph Anthony Ignagni In the first Establishment Clause case decided during Warren Burger's tenure as Chief Justice, the 0.3. Supreme Court laid down a new constitutional test. With this addi- tion, the Court now had in place the third prong of a three-part Establishment Clause test which continues to be used. However, this three-part test has not settled what is allowable in church-state relations for many scholars. In fact, it is often complained that constitutional law in this area is confused and conflicting. This dissertation attempts to show that these decisions are not as certain or unpredictable as has been previously claimed. A fact- attitudinal model is derived from judicial behavior theory, cybernetic decision making theory, and the writ- ings of the justices themselves. Probit is used to estimate its parameters. The results suggest that the model has explanatory as well as predictive value, and thus these decisions may' be far more consistent than scholarly opinion would have led us to believe. To my wife and family iv ACKNOWLEDGMENTS I wish to thank Harold J. Spaeth not only for his guidance in this project, but for introducing me to this entire field of study. I am also greatly indebted to David Rohde and the other members of my committee for their sup— port and advice. Lastly, I would like to thank Michael Keefe, Me-Young Ju, Timothy Hagle, and James Meernik for their help and encouragement at various stages along the way. TABLE OF CONTENTS List of Tables ......................................... vii Introduction ............................................. 1 1. Literature Review ..................................... 6 A. Traditional Legal Scholarship.. ................... 6 B. Judicial Behavior and Attitudinal Research ....... 10 C. Fact or Cue Models.... ........................... 16 2. The Justices of the Supreme Court as Human Decision Makers.... ....................... 21 A. Rohde - Spaeth Framework ......................... 21 B. Human Limitations and Bounded Rationality ........ 24 C. Cognitive - Cybernatic Theory .................... 29 D. The U.S. Supreme Court... ..... . .................. 4O 3. A Theoretical Perspective on the Burger Court’s Establishment Clause Decisions.... ....... . .......... 47 A. Overview ...... ... ..... ...... ..................... 47 B. Establishment Clause Typology .................... 56 C. Potential Complicating Issues .................... 76 4. Operationalization and Methodology ................... 82 A. Operationalization / Measurement, Model, and Data82 B. Methodology ...................................... 98 5. Presentation of Results.. ........................... 108 A. The Votes of all of the Individual Justices ..... 108 B. Reduced Model ................................... 121 C. The Justices One by One... ...................... 124 D. Decisions by the Court .......................... 135 6. Summary and Conclusion .............................. 139 Notes .................................................. 148 References ............................................. 150 vi LIST OF TABLES Table 1 ................................................ 110 Table 2 . .. . ........... . ............................ 114 Table 3 ................................................ 119 Table 4 ................................................ 119 Table 5 ................................................ 122 Table 6 ................................................ 123 Table 7 ................................................ 127 Table 8 ................................................ 127 Table 9 ................................................ 128 Table 10 ............................................... 128 Table 11 ............................................... 129 Table12 ................................................ 129 Table 13 ............................................... 130 Table 14 ............................................... 130 Table 15 ......... . ..................................... 131 Table 16 ............................................... 137 Table 17.. .............. . .............................. 137 vii Introduction Alexis de Tocqueville stated in 1835 that "scarcely any political question arises in the United States that is not resolved, sooner or later, into a judicial question"(1945,p.280). This certainly holds true in the area of church-state relations. The United States Supreme Court in the last 50 years has been asked repeatedly to interpret and then reinterpret the two clauses of the First Amendment which deal with the area of religion. The amendment mandates that "Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof. . ." While these clauses may appear to be straightforward at first glance, upon further reflection one realizes that they allow for numerous interpretations and at times may conflict with one another. These differing interpretations have led to a number of bitter disputes over the years, often in the area of education. As Martha McCarthy has noted,"The relationship between religion and government has created extensive controversy in the history of this nation ."(1983,p.1) It was in 1947 when the Supreme Court first found it necessary to define the meaning of the Establishment Clause. In Everson v. Board of Education, Justice Black wrote: The "establishment of religion" clause of the First Amendment means at least this. 1 2 Neither a State nor the Federal Government can set up a church. Neither can pass laws which aid religion, aid all religions, or prefer one religion over another. Neither can force nor influence a person to go to or remain away from church against his will or force him to profess a belief or disbelief in any religion. No person can be punished for entertaining or professing religious beliefs or disbeliefs for church attendance or non-attendance. No tax in any amount large or small, can be levied to support any religious activities or institutions, whatever they may be called, or whatever form they may adopt to teach or practice religion. Neither State nor the Federal Government can, openly or secretly, participate in the affairs of any religious organizations or groups and vice versa. In the words of Jefferson, the clause against establishment of religion by law was intended to erect "a wall of separation" between church and state (1947, p.15). While Black’s statement is probably the most often quoted language concerning the Establishment Clause, it by no means settled its exact meaning. What precisely breached this "wall of separation" between church and state continued to crop up. The Everson no-aid test was quoted and applied in several later decisions, however, in the early 1960's the Court enunciated a new test. In cases dealing with prayers and Bible reading in public schools, the Court used the "secular purpose and primary effect" test in making their decisions (Enqel v. Vitale, 1962; School District v. Schempp, 1963). This required that when a law was challenged under the Establishment Clause it must have both a secular purpose and a primary secular effect. Then in 19701, in the first Establishment Clause case decided during Warren Burger’s tenure as Chief 3 dimension of the present Establishment Clause test, the purpose-effect-entanglement test (Walg v. Tax Commission, 1970). This new test added to the purpose and effect test the requirement that a law must not involve the government in an excessive entanglement with religion. Lastly, in decisions from 1971-1975, the Court began placing increased reliance on whether the program being considered caused political division along religious lines (Lemon 2; Kurtzman, 1971; Meek v. Pittenger, 1975). The opinions in these cases have not made clear whether "political divisiveness" is a fourth factor to the test or merely part of the third prong. This series of tests or factors to be considered has not cleared up the matter for many scholars. In fact, it is often complained that constitutional law in this area is "confused, conflicting and uncertain"(Pfeffer, 1979, p.5). In the area of aid to nonpublic schools, McCarthy claims that the Supreme Court has not provided clear guidance, but rather has provided more questions than answers (1983, p.117). Jesse Choper has referred to these decisions as "ad hoc judgments which are incapable of being reconciled on any principled basis"(1980, p.680). This "confusion" has led some legal scholars to doubt that reasonable predictions can be made about future cases (Pfeffer, 1984, p.37). Evan Tager has written that "Establishment Clause cases have become totally unpredictable"(1984, p.235). John Nowak, Ronald Rotunda, 4 and J. Nelson Young have simply stated that "it would be foolish to predict the results of future cases"(1978, p.858). This dissertation will focus on the U.S. Supreme Court and its rulings involving the Establishment Clause of the First Amendment. It will attempt to show that Supreme Court decisions in this area are not as "conflicting" and "confused" as many scholars believe. The years to be analyzed will be those of the Burger Court (1969 through 1985 terms). This time period offers an interesting opportunity for which to examine the Court’s decision making. There were only thirteen justices (including Burger) who voted in these cases during these years. Furthermore, the Court created and purported to use the purpose-effect-entanglement test throughout the period. Thus, the relative stability in Court membership and constitutional doctrine allows one to study whether the individual justices and the Court behaved in a consistent manner. With these ideas in mind, this dissertation begins with a review of the previous work done in the field. It then moves on to theoretical development and the formation of a model. In addition, there will be a discussion of how the model can be operationalized and its parameters estimated using Probit. Both the decisions of the individual justices and the Court as a whole will be considered and tested. This model can hopefully not only explain and correctly 5 categorize previous Establishment Clause cases, but also be used in the prediction of future cases. Therefore, this research should contrast quite sharply with the work previously done in this field. Chapter 1: Literature Review "What I mean by law is nothing more or less than the prediction of what a court will do."--Justice Oliver Wendell Holmes A. Traditional Legal Scholarship Most of the work in this area has been done by tradi- tional legal scholars. This literature has tended to be qualitative and descriptive in nature. Often these works give an account of and analyze what the Supreme Court has done in particular cases. However, this research varies greatly in terms of the time span considered. First there are articles and research in this genre which discuss a recent landmark decision. For example, Wilber Katz interpreted the Supreme Court work in Walz v. I§§__Commission (1970) as dismantling' the old "leaning tower of absolutes" concerning the Establishment Clause and "presenting a rough sketch for a new edifice" (1975, p. 100). Donald Giannella found in reviewing the decision in Lemon v. Kurtzman (1971) and Tilton v. Richardson (1971) that the Court's reasoning "afforded little light on the constitutional status of the state aids and sub- sidies different from those directly at the issue" (1975, p.114). He further claimed that if the excessive entangle- ment test is to be taken seriously "it raises more ques- tions than it answers" and "another round of controversy appears inevitable" in this area of the law (p. 114-115). William Van Alstyne (1984) criticized the Court’s decision in Lynch v. Donnelly (1984). In dramatic fashion he wrote: 6 "I do not know whether Mr. Jefferson would have been sur- prised, but I believe he would have been disappointed" (p. 787). There are many other examples of this type of research including: Valenti (1969); Monaghan and Ariens (1984); Fairchild (1985); Mawdsley (1986);Reinertsen and Vinson (1986); and Smith (1986). Second, there is traditional legal research and analyses which assess the last few years or decades of Supreme Court rulings. A case in point would be Herman Schwartz’s review of the Burger Court’s Establishment Clause decisions (1987). Schwartz concluded that "while the path has been irregular, and the results far from coherent, a slender majority still appears to cling to constitutional principles that were first articulated by the court nearly four decades ago" (1987, p. 91). Another instance of this type of work was done by Henry Abraham (1987). His assessment of the Burger years is that "there is little doubt that a degree of erosion has taken place in the separationist commitment of the Court... "(1987, p. 37). Other examples of work of this kind include: Gaffney (1980; Pfeffer (1980); Ripple (1980a); Gemmer (1982); and Van Patten (1983). Third, there are also numerous books and articles which consider hundreds of years of church-state rela— tions. Often this research focuses on the Founding Fathers. This point can be illustrated by the ongoing debate concerning the meaning of the Establishment Clause at the time of the framing and adoption. A number of authors including Walter Berns (1976), Michael Malbin (1978), and Robert Cord (1982), have written that the Supreme Court has relied upon a flawed reading of the intentions of the authors of the first amendment. It is these scholars belief that the Establishment Clause was not intended to prohibit government aid to all religions or to religion on a nonpreferential or exclusive treat- ment.. Other schools have attacked this "nonpreferential" or accommodationist" interpretation (e.g. Pfeffer, 1987; Levy, 1986). Their reading of history is that any finan- cial aid to religion constitutes an establishment of reli- gion. Other books concerning the long term history of this area of law include: Kauper (1964); Howe (1965); Miller (1986); Bradley (1987); and Alley (1988). Law review articles of this genre often discuss a recent landmark decision (e.g., Valente, 1969; Van Alstyne, 1984; Mawdsley, 1986), or assess the last few years of Supreme Court rulings (e.g., Ripple, 1980a; Gaffney, 1980; Pfeffer, 1980; Gemmer, 1982; Van Patten, 1983). There are also numerous books which consider hun- dreds of years of church-state relations (e.g., Kauper, 1964; Howe, 1965; Pfeffer, 1967; Cord, 1982; Levy, 1986). In addition to research which gives an historical review and analysis of what has occurred in the past, some legal scholars have advanced different theories about how the Supreme Court should interpret the Establishment Clause, or how it should be defined. Philip Kurland (1961) has espoused a neutrality theory which states that government cannot impose penalties, or give favors, if the basis for doing either of these is religion. Jesse Choper has argued that "the Establishment Clause should forbid only government action whose purpose is solely religious and that is likely to impair religious freedom"(1980, p.675). One commentator has observed that the Court could put an end to its conceptual difficulties if it would return to a no-aid position since this is "a guiding prin- ciple that possesses functional integrity"(Buchanan, 1978, p.835). Another scholar has written that the Court should change or eliminate the effect test and instead erect a stopping point, a point beyond which government aid is prohibited (Teger, p.237). This qualitative research provides a needed back- ground on what the Court has done, and what facts or fac- tors may affect Supreme Court decisions. However, in addition to this literature, one also needs to review a quite distinct area of judicial research. This second area involves more general theory concerning the behavior of judges or justices. This area studies such topics as the values, attitudes, and perceptions of those behind the bench, and how these affect decision making. In this 10 endeavor, quantitative and statistical techniques are often relied upon. B. Judicial Behavior and Attitudinal Research Prior to World War II, the Court was primarily viewed as a passive institution. A justice's job was not to make law but to "discover" what the law was. A justice would "find" the law and then apply it to the case before him. Justice Roberts claimed in United States v. Butler that the Court had a simple duty when considering whether a particular law was constitutional or not. The justices merely had "to lay the article of the Constitution which is invoked beside the statute which is challenged and to decide whether the latter squares with the former" (1936, p.62). This view of the Court has sometimes been called "mechanical jurisprudence," "declaratory theory," or the "Cult of the Robe" (Goldman and Jahnige, 1976, p.155; Spaeth, 1979, p.3; O'Brien, 1986, p.44). Members of the Court simply declare what the law is, in a neutral and objective fashion, without recourse to, or interference from, their own life experiences and personal values. They have little or no discretion. Whatever they may once have learned or believed, when they don the robe justices become detached servants of the law. They have no will, only judgment (Grossman and Wells, 1988, p.86). When this concept of jurisprudence was taken to an extreme, it took on mythological and religious imagery. 11 Justices were seen as Delphic oracles. They were not fal- lible human beings giving their opinion, but rather the Constitution and laws were speaking through them (Spaeth, 1979, p.2). Justice Frankfurter claimed that "when a priest enters a monastery, he must leave-or ought to leave-all sorts of worldly desires behind him. And this Court has no excuse for being unless it's a monastery" (O’Brien, 1986, p.84). Anthony Lewis has written that the Court’s public image "seems sometimes to be less that of a court than of an extraordinary powerful demigod sitting on a remote throne and letting loose constitutional thunder- bolts whenever it sees a wrong crying for correction" (1964, pp. 11-12). This mystique surrounding the Court began to dis- sipate once researchers started to study its actions in a more systematic and scientific fashion. The first major break from past notions of the behavior of the justices came with C. Herman Pritchett’s work in the 1940’s which included his book The Roosevelt Court: A Study in Judicial Politics and Valugs. 1937-1947 (1948). Pritchett states: I began to wonder what it was in that case and in the autobiographies of those justices that led them to disagree with the majority of the Court on the issue there raised. I decided that it might be profitable to examine into the actual patterns of dis- agreement among the justices...(p.xi) Pritchett's work was seminal for two reasons. First, it characterized the Supreme Court as a small decision making 12 group, made up of individuals who vote in accordance with their attitudes about different issues. Second, the research involved the quantitative analysis of a large sample of data. This laid the groundwork for other pioneers in the field of judicial behavior or judicial politics, including Glendon Schubert. Among Schubert’s significant contribu- tions to this area of study, was his book Quantitativg Ag§1y§i§_of Judicial Behavior (1959). This work can almost be considered a "how to" book or as a training guide in the use of several quantitative research techni- ques. C. Neal Tate writes, "The most substantial influence on the use of quantitative methods in judicial behavior came initially from Quantitative Analysi§_gf Judicial Behavior"(1983, p.71). I These early works helped lead to a boom in judicial behavior research in the early 1960's (e.g. Ulmer, 1960, 1961a, 1962; Tanenhaus, 1961; Hayakawa, 1962; Spaeth, 1961, 1962, 1963a, 1963b, 1964; Schubert, 1961, 1962, 1964, 1965). The research was often quantitative, and judges were no longer viewed as impartial or detached. Instead courts, especially the U.S. Supreme Court, were treated as political agencies and judges as political actors. For example, Walter Murphy wrote in 1964 that Justices of the Supreme Court had the capability through 13 the peculiar kinds of authority and discretion inherent in their position to develop public policy (p.2). Harold Spaeth has elaborated on this point by stating that the Court’s rules "do not preclude any justice from voting compatibly with his personal policy preferences"(1979, p.11). Justices are not electorally accountable,they generally lack ambition for higher office, and the Supreme Court is the court of last resort. Thus, in most situa- tions "the Justices are free to decide cases as they see fit"(p.118). Closely tied to these ideas was the development of attitudinal models to describe the behavior on the Supreme Court. Since the justices have few decision making con- straints placed upon them, the question became: what does determine how they act? Several scholars answered that it was the justices’ substantive attitudes and values which greatly affected their votes in cases (e.g. Pritchett, 1941, 1948; Schubert, 1961, 1965; Tanenhaus, 1961; Spaeth, 1963a, 1963b, 1979). Pritchett wrote: "Private attitudes, in other words, become public law" (1941, p.890). Supreme Court decisions are consistent, not because they flow from precedents, statutes, or the U.S. Constitu- tion, but because of the consistency with which justices follow their own values and biases (Goldman, 1979). There is "a kind of stare decisis underlying the Supreme Court's decisions but it is based on personal rather than institu- l4 tional precedents" (Schubert, 1974, p.20; see also Lawlor, 1967; Shapiro, 1972; Flango and Ducat, 1977). This per- sonal consistency allowed researchers to use cumulative scaling techniques in an attempt to uncover the values and decisional patterns of the justices. David Rohde and Harold Spaeth, using cumulative scales, found there to be three underlying values which explain most Supreme Court decisions: freedom, equality, and New Deal economics (1976, pp.137-138). Furthermore, they were able to predict 86% of the votes of the individual justices they analyzed, and almost 88% of the outcomes of the cases under study (p.157). While attitudinal models have led to great progress in the ability to explain and predict behavior on the Supreme Court, such models have been criticized. Several traditional public law scholars attacked this new approach to the study of courts and judges (e.g. Roche, 1958; Bec- ker,1963; Mendelson, 1963, 1964a, 1964b, 1966). One of the main criticisms was that this new type of research did not sufficiently capture the complexity of the judicial process. Later, some scholars claimed that scaling had certain limitations (Tanenhaus, 1966; Gibson, 1978, 1983; Tate, 1983). It was argued that when scales are used, subjective judgments must be made by the researcher. Two individuals with the same data may predict different out- comes. Another criticism was that attitude models were 15 unidimensional. Judicial decision making analysis ought to be multidimensional (Tate, p.72). Those involved in this research have repeatedly responded to these charges (e.g. Schubert, 1958, 1963, 1967; Ulmer 1961b, 1969; Kort, 1964a; Spaeth, 1965; Spaeth and Peterson, 1971). They point out the advantages in going from a subjective and impressionistic mode of analy- sis to systematic and replicable studies. In addition, there have been major strides in the complexity and multi- dimensionality of models of Court action (e.g. Schubert 1965, 1974; Rohde and Spaeth, 1976; Spaeth, 1979). Gibson admits that "there is little question that the predominant paradigm of judicial decision making places judges’ atti- tudes in the center of the process. Indeed, it is not an overstatement to assert that attitudinal approaches have become the traditional nontraditional mode of judicial analysis" (1978, p.912). Sheldon Goldman and Austin Sarat conclude that "it is fair to observe that the continued finding of patterns of voting (by Supreme Court justices, and lower court judges) that can be interpreted as representing attitudes and values puts the overwhelming weight of evidence on the side of the attitudinalists" (1989, p.386). 16 C. Fact or Cue Models This brings us to the final area of judicial litera- ture which should be reviewed for the purposes of this study; research dealing with the influence of case stimuli. A number of researchers, over the years, have stressed the importance of the particular facts or cues in the cases which appear before the Supreme Court. James Gibson writes: "Attitudes alone cannot determine behavior; before attitudes even become relevant, they must be stimu- lated" (1983, p.13). One of the earliest nontraditional works focusing on the facts of cases was done by Fred Kort (1957). It was an ambitious attempt at predicting Supreme Court decisions mathematically. Kort stated his study was designed to demonstrate that it is possible to take some decided cases, to identify factual elements that influenced the decisions, to derive numerical values for these elements by using a formula, and then to predict correctly the decisions of the remaining cases in the area specified (ID-1)- More specifically, Kort analyzed the content of the Court’s opinions in the "right to counsel" cases from 1932 to 1947. He arrived at a list of twenty-six "pivotal fac- tors" (p.4). Kort then divided the data into two chronologically ordered sets of fourteen cases. He created a formula to assign a weight to each factor so that when the weights were added for all the factors pre- sent in a specific case, the resulting sum was greater in 17 all cases decided in favor of the defendant than in any case decided against the defendant. Once this weighting scheme was determined based on the first fourteen cases (the source group), he then applied it to the later group (the test group). Kort was then able to successfully predict twelve out of the fourteen test group decisions, with two cases lying in a zone of uncertainty (p.11). A number of studies were to follow this initial work. Schubert applied Kort's method to a parallel set of search-and-seizure cases (1959). Both Stuart Nagel and S. Sidney Ulmer proposed simpler systems of weighting factors as substitutes for Kort’s approach (1960; 1962). Kort himself then attempted a more sophisticated weighting scheme based upon factor and regression analysis (1963). Kort and Reed Lawlor experimented with the use of logic and Boolean algebra (1964b; 1964). Discriminant analysis was applied to the search-and-seizure cases by Ulmer (1964). Kort continued his investigation of the connec- tion between fact patterns and court decisions and eventually explored the possibility of a nonlinear rela- tionship existing (1966, 1968, 1973). Lastly, there have also been successful uses of fact models in the 1980’s (e.g. Ulmer, 1981; Segal, 1984, 1985, 1986; Gryski, Main, and Dixon, 1986). While this research has consistently been able to achieve high levels of prediction, much of it has been 18 flawed in terms of methodology and research design. Some of the early works had more variables than cases. Many of the weighting schemes told little about the importance or significance of various factors. Changes in court person- nel were usually treated as making no difference in the outcomes of cases. For example, in Kort’s first study, he ignored the changing composition of the Supreme Court even though twenty-five justices sat during the time period analyzed. Until recently, fact models had simply left the study of individual justices untouched. Only decisions of entire courts had been considered. Finally, it should also be noted, that with few exceptions (e.g. Kort, 1973; Segal 1984, 1985, 1986), there has been no theoretical justification for the methodology used. Different for- mulas and methods have been tried with no clear reasons given. This brings up the second and more significant fail- ing of this research; the lack of theoretical development. While later work involving fact models improved meth- odologically, little attention was paid to theory (with the possible exception of Segal’s research). Many of the writings simply indicate that a certain combination of facts lead to decisions in favor of one party while there are other combinations that lead to decisions in favor of the opposing party. Why is this? What theory of decision making underlies such a proposition? 19 Schubert was critical of this void in theory (1964). He pointed out that with no theoretical explanations given, the work of Kort and others could be viewed as merely a special variant of mechanical jurisprudence. Facts could be seen as determining or controlling the decisions of judges (p.451). He, of course, felt this was inappropriate. Instead, judges dominate and control facts, not vice-versa. Schubert did not reject this research outright, but simply felt it was not being properly approached theoretically. He writes that "from a behavioral point of View, to study the facts in cases is to study attitudinal data" (p.452). What are deemed to be significant or pivotal facts in a case relate back to the perceptual processes, attitudes, and values of the judges involved. The facts are interpreted by a judge in accord- ance with his or her values and personal decision cal- culus. Thus, these types of studies could be treated as falling nicely in line with research done by those in judicial politics and behavior. One final line of inquiry concerning the importance of facts needs to be discussed. Research dealing with case stimuli was taken in slightly different direction by Tanenhaus et al. (1963) with their work on "cue theory". Their theory (dealing with the Court’s certiorari juris- diction) basically states that since certiorari petitions are so sizable and numerous, justices can give no more 20 than cursory attention to a large share of them. There- fore, there must exist a group of readily identifiable cues which enable the justices to separate "the certiorari petitions requiring serious attention from those that are so frivolous as to be unworthy of careful study" (p.118). They tested for the presence of four cues and hypothesized that, if one or more of the four cues were present, then the likelihood of certiorari being granted would be greater than when no cues were present. Three of the four hypothesized cues passed their test. They concluded that the federal government as petitioning party, dissension among judges or lower courts, and the presence of a civil liberty question were cues promoting the granting of the writ. Other judicial scholars have followed Tanenhaus and his associates in stressing that due to constraints of time and resources, cues which alert a justice to the pos- sible importance of a case could be very useful (e.g. Ulmer et al. 1972; Songer, 1979; Armstrong and Johnson, 1982; Ulmer, 1984). Yet, writers including Stuart Teger and Douglas Kosinski (1980) have pointed out that there are some prob- lems with the previous work done on cue theory. They men- tion that since the cues chosen are surrogates for salient issues, they may have to be updated. (It should be noted, however, that Armstrong and Johnson's study in 1982 indi- cated that "the Tanenhaus cues are alive and well in the 21 United States Supreme Court" [p.150].) In addition, the way in which the theory has been tested borders on cir- cularity and nonfalsifiability (Gibson, 1983, p.14). All cue theory requires is that the presence of a cue is enough to insure that a petition will be studied carefully-not that certiorari will be granted. This is never directly tested. "It should be obvious that testing the validity of cue theory is almost impossible. A direct test would require access to the Justice's chambers, pre- ferably while he is shifting through the petitions for certiorari" (Teger and Kosinski, 1980, p.836). While it appears quite credible that case stimuli influence decisions, "little theory has been constructed that specifically details the processes of influence"(Gibson, 1983, p.13). Likewise, "Cue theory has always been more intuitively attractive than theoretically well developed or empirically well supported"(p.14). When considering the number of cases which come before the Court, in one form or another, it simply sounds reasonable that decision making shortcuts are developed. However, no cognitive theory underlies cue theory, and the statistical support for it is somewhat suspect. In conclusion, each area of literature which has been discussed, has produced some positive results. Yet, there are also major gaps or problems with what has been done. The writings by traditional legal scholars are helpful in 22 a descriptive sense, and can possibly be used in conjunc- tion with cue theory or fact pattern analysis to point out which factors are the ones which influence the justices. However, this work is qualitative and often normative, and thus, by itself, has limited value. Fact models and cue theory have intuitive appeal, but further theoretical and statistical proof needs to be shown. Lastly, while atti- tudinal models have been quite impressive in their ability to predict Supreme Court decisions, they would possibly be of even greater value if combined with other approaches. This study is an attempt to tie, to some extent, these different scholarly works together, and possibly fill in some of the gaps which exist. The next chapter of this dissertation suggests the integration of cue or fact theory with attitudinal theories. It also provides a more elaborate theoretical basis for cue theory. As will be seen, the integration and further development of this previous work can hopefully be used to more fully explain and predict the Supreme Court's behavior in Establishment Clause cases. Chapter 2: The Justices of the Supreme Court as Human Decision Makers "For few are likely to deny that justices of the Supreme Court have always to paraphrase Justice Frankfurter, ’read the laws of Con- gress through the distorting lenses’ ground by their own experience."--C. Herman Pritchett The review of the literature, pertaining to this dis- sertation, indicates that the research can be divided into three distinct groups. There has been descriptive and normative work about Supreme Court decisions in Estab- lishment Clause cases. Scholars have developed atti- tudinal models concerning the justices and their behavior. Lastly, there has been work which has concentrated upon the factual stimuli before the Court. However, no one has attempted to combine these bodies of research into a more coherent whole. In addition, many of these writings lack a theoretical foundation. This chapter will address these issues, and in doing so, attempt to make theoretical pro- gress. A. Rohde-Spaeth Framework David Rohde and Harold Spaeth have written that Supreme Court decisions "are the consequence of three fac- tors: goals, rules, and situations"(1976, p.70). These three factors have also been used as a framework for explaining all political decisions (Aldrich, Miller, 22a 22b Ostrom, and Rohde, 1986, p.19). Rohde and Spaeth assume that the justices have certain goals they wish to achieve. In their decision making on the Supreme Court, these goals are policy goals. "Each member of the Court has prefer— ences concerning the policy questions faced by the Court, and when the justices make decisions they want the out- comes to approximate as nearly as possible those policy preferences"(Rohde and Spaeth, p.72). It should be pointed out that these personal policy preferences are based on the individual’s beliefs, attitudes, and values (pp.75-78). Thus, the justices have certain policy goals which are based on their policy preferences, which in turn spring from their attitudes and values. While each individual’s goals and personal policy preferences matter a great deal, they do not operate in a vacuum. In addition to goals and preferences, a justice’s decision could be affected by the "rules of the game", the second factor in the Rohde-Spaeth framework. "These rules of the game, or rule structures, are the various formal and informal rules and norms within the framework of which decisions are made"(p.70). In the case of the Supreme Court, there are a number of rules which influence how it operates. For example, the Court must wait for actual cases to come to it, and cannot give advisory opinions (Ulmer, 1986, p.13). Before a case is granted review, four of the nine justices ("The Rule of Four") must be in 23 favor of hearing and deciding the case (p.16). The Court also has no means by which to enforce the decisions it hands down, but must instead rely upon the executive branch (O’Brien, 1986, pp.317-320). Yet, while there are certain restrictions placed upon the Supreme Court, the rules of the game also allow the justices great liberty in their actions. As stated in the last chapter, the justices are not electorally account- able, they generally lack ambition for higher office, and the Supreme Court is the court of last resort. Therefore, the members of the Court are relieved of many of the pres- sures felt by policy makers in other branches of the government. Instead, the rules give the justices flexibility to pursue their goals. Rohde and Spaeth con- clude "The Supreme Court’s rule structure permits the jus- tices greater freedom than other political decision makers to base their decisions solely upon personal policy pref- erences..."(1976, p.72). The third factor in this framework is the particular situation facing the Court. In addition to goals and rules, "decisions are affected by the particular con- figuration of circumstances that constitute the various decision-making situations in which individuals find them- selves"(p. xv). Thus, the specific situational factors or facts in a case before the Court can influence a justice’s 24 vote. If the facts or situation change, so can the deci- sion of the Court. This third factor is, of course, closely related to the ideas underlying fact models and cue theory. It is also the linchpin for what is to be discussed next. For in the remainder of this chapter there will be an attempt to fuse previous attitudinal research with cue theory and fact models (which indirectly also incorporates some of the descriptive work which has been done). This will be accomplished by applying Herbert Simon’s view of decision making and cognitive-cybernetic theory. In doing so, the framework given by Rohde and Spaeth will be elaborated upon and expanded. Their work dealt primarily with the first step of the framework (due to the freedom allowed by the rules of the game--step two). The third step was not considered to the same extent. In addition, as discussed above, cue theory and previous fact models have lacked theoretical support. Therefore, this theoretical discus- sion will focus upon the importance of facts or cue, and then will be tied back to the other two parts of the framework. Lastly, the Supreme Court will be examined in terms of these concepts and ideas. B. Human Limitations and Bounded Rationality There are different views of how human decision makers behave. In this work, human decision makers 25 (Supreme Court justices) will be viewed in terms of con- strained maximization, and cognitive-cybernetic theory; thus combining the work of Herbert Simon and John Stein- bruner among others. If individuals had unlimited com- putational powers and resources they might behave as util- ity maximizers, but it will be assumed here that instead their behavior is boundedly rational. Simon writes that if one takes into account the limitations of knowledge and computing power of a choosing organism, he or she may find it incapable of making objec- tively optimal choices (1985, p.294). However, if this organism uses methods of choice which are as effective as its decision making and problem-solving permit, then one could speak of bounded rationality, that is, "behavior that is adaptive within the constraints imposed both by the external situation and by the capacities of the deci- sion maker" (p.294). More specifically, bounded rationality is defined by Simon as rationality in situa- tions where the complexity of the environment is immensely greater than the computational powers of the adaptive system (1981, p.190). Relating this to humans, he states: The capacity of the human mind for for- mulating and solving complex problems is very small compared with the size of the problem whose solution is required for objectively rational behavior in the real world--or even for a reasonable approxi- mation of such objective rationality (1957, p.198). 26 Therefore, according to Simon, if humans are to behave rationally, the most they can achieve is bounded rationality. There are two reasons for this. First, there is the complexity of the world. Complexity or a complex system is "made up of a large number of parts that interact in a nonsimple way"(1981, p.195). Steinbruner claims that "complexity seems to describe both the government and contemporary social conditions..."(p.15). In and of itself, complexity is not a problem. Simon and Christopher Alexander have written that complexity can be handled if it is broken down into its simpler subparts (Simon, 1981; Alexander, 1964). However, this brings us to the second reason why human behavior is quite often boundedly rational; the limitations of the human mind. For over thirty years Simon has focused closely upon the limits of human rationality and information process- ing. In his research on memory structures, learning processes and problem solving, he has stated repeatedly that man has severe limitations upon what he is able to mentally accomplish (Simon, 1979). For example, Simon asserts that humans’ short-term memory structure has a very small capacity, and it requires a relatively long time to transfer a chunk of information to long-term memory (1981, p.96). These limitations result in the com- plexity of the environment often being greater than the computational powers of the human mind. Simon concludes: 27 "A man viewed as a behaving system is quite simple. The apparent complexity of his behavior over time is largely a reflection of the complexity in which he finds him- self"(1981, p.65). Faced with such limitations in a complex world, how do individuals make decisions? Simon writes, "A real-life decision involves some goals or values, some facts about the environment, and some inferences drawn from the values and facts"(1959, p.273). Before going into more detail, it should be noted that there are definite parallels between this decision making model and the Rohde-Spaeth framework for Supreme Court decision making. Both stress the importance of values and the particular factual situa- tion involved. According to Simon’s model, individuals have certain goals and values. With this as a starting point, they observe their environment. However, the information and facts which are fed into the person’s cognitive structure may not be an accurate approximation of the real environ- ment. The individual’s perception acts as a biased filter in determining what is included and excluded from con- sideration. There are simply more bits of information than a human being can possibly absorb or retain, and so information is omitted (1959, p.273). "People are, at best, rational in terms of what they are aware of, and 28 they can be aware of only tiny, disjointed facets of reality" (1985, p.302). The individual then uses his or her values and the perceived facts as premises. Simon states that "the deci- sion which is finally reached is inferred from these premises"(1959, p.273). Thus, he refers to them as deci- sion premises. They can be viewed as instructions or "computational procedures for assessing the state of the environment and its implications for action..."(p.274). He claims that the whole procedure can be viewed as a process of reasoning (p.273). While not a strict form of logical reasoning, this process does have a certain struc- ture. The individual’s values and their perceived facts about the environment are used as premises. These premises can be seen as being decision rules or instruc- tions for what should be done. he individual then draws certain inferences from these premises and acts accor- dingly. Closely tied to this process of decision making is the notion of "satisficing". As has already been dis- cussed, due to uncertainty, complexity, and limitations on human knowledge and ability, people cannot always maximize their utility. Faced with this situation, Simon argues, individuals "must be content to satisfice--to find ’good enough’ solutions to their problems and ’good enough’ courses of action"(1979, p.3). The claim is that instead 29 of perfect or optimal solutions, individuals are often content to achieve satisfactory ones. They have some specified level of basic satisfaction which is below their optimal level (p.26). Satisficing is a relatively simple mechanism for decision making. It does not require com- prehensive knowledge or comparisons because it has an end of search rule. The search ends when a good enough alter- native is found (p.3). Therefore, Simon believes that man is a satisficer, "not because he prefers less to more but because he has no choice"(1981, p.36). So, when moving from the decision premises to making a decision, Simon would not expect maximizing but rather a satisficing decisional process. The inferences made and options considered would be quite limited. In fact, the options were already greatly reduced by the person’s values, goals, and perceptions. Only a few of the pos- sibilities made it through this first screening device. Thus, it is quite likely the best or most optimal alterna- tive will not be chosen, but instead only a satisfactory one. However, in conclusion, Simon claims that it is incorrect to call this "irrational" behavior; instead it is better viewed as a form of rationality (1985, p.297). C. Cognitive-Cybernetic Theory Similar in many ways to the work of Simon is the field of cybernetics. Cybernetics also provides a view of 30 decision making which is far simpler than other more com- prehensive rational models, but which can nonetheless be "highly successful in the proper environment"(Steinbruner, p.13). It should be pointed out that there are a number of versions of the cybernetic model (e.g., Wiener, 1948; Ashby, 1956; Beer, 1964; Steinbruner, 1974). For the pur- poses of simplification and relevancy, Steinbruner’s work will be primarily relied upon. Let us begin our discussion of the cybernetic model (as Steinbruner does, p.49) with what would appear to be a relatively simple decision making situation: a person playing tennis. At the most basic level, a tennis player attempts to serve and hit ground strokes to appropriate areas of the opposing player’s side of the court. However, when analyzed more closely, the activity can be viewed as being far more complex. For example, when receiving the ball, it flies across the net at different speeds and trajectories. The player must decide where to intercept the ball, what stroke to use, how hard to hit it, and where to place it in the opposing court. As part of these decisions, the tennis player must decide whether to move right, left, forward, or backward, and at what angle. He or she must decide how quickly to move toward the ball. In what grip or position should the racket be in, and at what angle should the ball be swung at? Where is the opposing player positioned? Upon hitting the ball, 31 should he or she charge the net, or retreat to the baseline? Without having exhausted all of the possible complexities of the game, the number of decisions is already rather large. In Steinbruner’s simplified version of a game of tennis, considering different sequences and combinations of decisions, for each stroke of the ball "there are 4,200 different solutions to the problem for the player"(p.50). It should be kept in mind that these decisions need to be made in a few seconds or less. Steinbruner cites the tennis player as a clear exam- ple of a situation at odds with any decision making model requiring comprehensive calculations or comparisons. He claims it is unrealistic to expect an elaborate or utility maximizing procedure to be relied upon. "The speed at which tennis players can find a sclution makes it very unlikely that they examine all of the alternatives, estab- lish a preference ordering, and choose the best one"(p. 50). Instead, Steinbruner argues, there must be a far more simple decision mechanism. Steinbruner and others have therefore looked to cybernetics, which can be viewed as an art of "steermanship"(Beer, p.30), and has been defined as "the science of control and communication, in animal and machine"(Ashby, p.1). Cybernetics rejects decision making requirements such as comprehensive information, cost-benefit calculations for all consequences, and optimizing comparisons. "The 32 cybernetic theorist doubts that decision makers engage in sophisticated outcome calculations with any degree of regularity or consistency"(Steinbruner, p.66). In fact, the central focus of cybernetics is to eliminate the vari- ety which is inherent in any significant decision (p.56). Instead of considering all of the possibilities and making the appropriate calculations, it attempts to control vari- ety and uncertainty by means of highly focused attention and highly programmed response. A cybernetic animal or machine relies on a simple decision rule or mechanism which allows for adaptive behavior and survival. To illustrate the cybernetic ideal of uncertainty and variety control, let us consider two man-made devices: the thermostat and Watt governor (sometimes referred to as "servomechanisms", Steinbruner, p.51). Thermostats are set at a certain desired temperature and record what the temperature actually is in a room or building. When the temperature deviates from the desired set temperature, a process is activated which then cools or heats the room or building back to the acceptable temperature level. Instead of regulating temperature, the Watt governor was devised to control the speed of a steam engine. The Watt governor was designed so that as the speed of the engine increases two steel balls on arms swing outward by centrifugal force and this closes the engines throttle. As the throttle closes, the speed of the engine decreases. 33 When the speed decreases, the steel balls fall downward because of gravity, and this once again opens the throttle and increases the speed of the engine. This process of opening and closing the throttle alternates back and forth (see Steinbruner for a more detailed discussion, pp.51- 52). Both of these devices have some basic underlying working principles in common. Both the thermostat and Watt governor focus on one primary or critical variable. In the case of the thermostat it is temperature, and for the Watt governor it is engine speed. Furthermore, both devices operate to keep the critical variable within a particular and acceptable range. Each reacts as the critical variable moves outside of the tolerable range in order to return it to the specified level. Lastly, each mechanism achieves the desired level of the critical vari- able by relying on a simple decision process. While devices such as thermostats and Watt governors can be seen as "extreme" or "pure" forms of cybernetic decision making, this model can also be applied in broader and more complicated settings. Its principles can be stretched from simple machines to human behavior. Accord- ing to Steinbruner, human beings attempt to hold the psychological effects of uncertainty to a minimum (p.66). A human decision maker is "engaged in buffering himself 34 against the overwhelming variety which inheres in his world..."(p.66) Thus, Steinbruner believes that uncertainty control occurs in the decision making process of human thought. Completely in conformity with Herbert Simon, Steinbruner holds that uncertainty control entails highly focused sensitivity (p.67). Individuals are not capable of accurately perceiving every feature of their environment. Some variables are focused upon, many others are left out. "The cybernetic thesis is that the decision mechanisms screen out information which the established set of responses are not programmed to accept"(p.67). In other words, uncertainty and variety are greatly reduced because only a few critical variables receive attention. The decision maker leaves most of the environment outside of the decision making process. In more specific terms, Steinbruner views the cybernetic decision making process progressing in the fol- lowing manner. A cybernetic decision maker focuses on a few incoming variables while eliminat- ing entirely any serious calculation of probable outcomes. The decision maker is assumed to have a small set of "responses" and decision rules which d e t e r m i n e t h e course of action to take once he has received information to which he is sensi- tive. That is, decision rules associate a given action with a range of "values" for the critical variables in f ocu s . . . The "responses" are action sequences,of the character of a recipe, established by prior experience. They are programs which accept 35 and adjust to very specific and very limited kinds of information (p.66). Once again, the work of Simon and Steinbruner match up quite nicely. Not only do both believe that decision makers omit information and instead focus on only a few variables, but both stress the importance of decision rules or premises. Upon tracking certain critical vari- ables, the decision maker does not make extensive calcula- tions nor a random decision. Rather, the individual operates in a characteristic fashion upon receipt of the perceptual input. He or she relies upon a set of estab- lished responses (assuming the individual has faced a similar situation previously). The decision maker con- siders the particular factors involved and the individ- ual’s course of action comes from his or her decision rule(s) for such a situation. Thus, in many ways like the thermostat or Watt governor, human decision making can possibly be viewed in quite simple terms. There is limited input (very few variables are considered), and the response to that input is also limited (prior experiences and decision rules are relied upon). Much of the complexity and uncertainty of the world is left untouched. The behavior of tennis players can be more easily explained in this context. Tennis player simply rely upon what they have done or seen before in similar situations. They have simple decision 36 rules which they continuously use (e.g., when at the net switch to the backhand grip). Tennis players do not have the time for more elaborate calculations. Therefore, in a world which is often quite complex, cybernetics can pro- vide decision makers with the "ability to produce strikingly adaptive outcomes in very complicated environ— ments" (Steinbruner, p.56). However, this overview of the cybernetic model would not be complete without at least a brief discussion of cognitive theory. Steinbruner believes that, at times, the cybernetic paradigm should be supplemented (p.14). He argues that simple cybernetic decision making has problems when the environment is not highly structured and when it is necessary to make inductive inferences (pp.13-14). Consequently, under these conditions, Steinbruner states that the process of uncertainty control depicted by the cybernetic paradigm needs to be amended (p.123). In those situations, "One must reach outside of the simple cybernetic paradigm for a more elaborate treatment of the high-level thought processes of the human mind"(p.87). Steinbruner’s answer to these problems is cognitive theory. There has been a recent outpouring of studies dealing with this field (e.g. George, 1980; Markus and Zajonc, 1985; Larson, 1985; Lau and Sears, 1986). In simple terms, cognitive theory explains how human beings structure their beliefs. "Cognitive principles offer an 37 analysis as to how highly complex decision problems are given the stable structure necessary for cybernetic processes to operate..."(Steinbruner, p.14). The mind is seen "as a mechanism for resolving ambiguity, as an inference machine which actively manipulates the informa- tion it receives..."(p.90). According to Steinbruner there are five basic princi- ples of cognitive operation: inferential memory, con— sistency, reality, simplicity, and stability (p.103). These principles account for the mind’s ability to impose structure on a person’s beliefs. In other words, the way in which relationships between beliefs are organized and how information is processed in reference to existing beliefs (p.95). These principles or structural regu- larities then explain how decisions can become structured. The first principle, inferential memory, involves "hierarchical" and "lateral" relationships in memory (p.95). In general, the content of memory is organized in a hierarchical fashion. Individuals tend to remember large overall concepts and are very loose with details. "Thus, if a person is asked to recall the Munich con— ference, he is very likely to think first of the notion of appeasement and then gradually bring back details--the specific issue of Czechoslovakia, the participants, the actual date"(p.96). 38 The second principle of cognitive theory holds that the mind is constrained by the structural regularity of consistency. "This simply means that the mind operates in such a way as to keep internal belief relationships con- sistent with one another, a constant which affects both the organization of memory and the processing of new information"(p.97). This cognitive principle has been repeatedly illustrated in experiments where the subjects’ perception of an object or event is distorted in order to conform to their past experiences or expectations. Third, the reality principle "asserts that the human mind is in contact with its environment, that stable, important features of the environment impose themselves quite reliably on the mind (p.100). This simply means that the mechanisms and operationsof the mind are often and in many key ways constrained by reality (p.101). Although, it should be kept in mind that an individual’s perception of the environment may, at times, be distorted or biased. This was just mentioned in regards to the con- sistency principle, and also above in terms of bounded rationality and cybernetic theory. In fact, this also relates to the fourth cognitive principle--simplicity. Steinbruner points out that the world or reality is enormously varied, complex, and chang- ing. The information processing capacities of the mind cannot handle recording everything which is experienced. 39 Instead, "the mind is highly selective about the informa— tion to which it attends about that which it uses. The mind remembers some things of importance but forgets a great deal and never even attends to most of the informa- tion it physically receives"(p.101). Thus, the principle of simplicity holds that cognitive mechanisms work to keep the structure of belief as simple as possible. In gen- eral, individuals will "establish simple belief structures rather than elaborate ones"(p.102). The fifth principle, stability, is tied to the fourth. Both are principles of economy. "The principle of stability asserts that cognitive inference mechanisms resist change in the core structures of belief"(p.102). These mechanisms are biased against change once a belief structure has been established, and so the system is kept as stable as possible. This ends the overview of cybernetic and cognitive theory. It should again be mentioned that in relation to the cybernetic paradigm, cognitive theory offers sup- plementary support. It should not be viewed as separate or distinct from the cybernetic model. In fact, the two overlap in many respects. Cognitive theory simply amends the cybernetic paradigm in certain situations. As stated above, it provides the stable structure necessary for cybernetic processes to operate. It therefore helps to 4O explain how highly complex decisions can be made by a simple decision making process. Cognitive theory also conforms to the work of Herbert Simon. In the following statement he captures and sum- marizes much of what has been asserted in this chapter. The human capabilities for rational behavior that are described by contemporary cognitive psychology are very congenial to the paradigm of bounded rationality...The models of problem solving describe a person who is limited in computational capacity, and who searched very selectively through large realms of possibilities in order to discover what alternatives of action are available, and what the consequences of each of these alternatives are. The search is incomplete, often inadequate, based on uncertain information and partial ignorance, and usually terminated with the discovery of satisfactory, not optimal courses of action (1985, p.295). D. The U.S. Supreme Court This theoretical view of decision making must now be related back to the U.S. Supreme Court and previous liter— ature on this institution. Henry Abraham has stated: "That the Supreme Court of the U.S. is a busy tribunal is axiomatic"(1987, p.33). Such a claim is based on two fea- tures of the Supreme Court. First of all, the Court often decides, what are usually considered to be, difficult policy and legal questions. Secondly, it can be argued that the Court is responsible for a very large amount of work. In fact, David O’Brien has asserted that "the 41 Court’s docket has grown phenomenally" in the last fifty years (1986, p.147). Other scholars writing on this change in the Court’s caseload have described it as a "massive increase", a "huge increase", "spectacular growth", and "dramatic growth"(Baum, 1985, p.101, 102; Grossman and Wells, 1988, p.46; Stumpf, 1988, p.133). As can be seen in Table 1, the October Term of 1980 had nearly five times the number of docketed cases as there were in 1930, and well over twice as many as in 1960. Justice Stevens has claimed that the justices are "too busy to decide whether there [is] anything [they] can do about the problem of being too busy" (O’Brien, p.153). In 1978, all nine members of the Burger Court signed a public letter which included the statement that "the Court’s caseload is heavy and growing"(Grossman and Wells, p.49). Table 1 Cases Docketed (U.S. Supreme Court) Fall Term Number 1930 1,039 1940 1,109 1950 1,335 1960 2,313 1970 4,212 1980 5.144 (Source: Stumpf, p.136) Let us take one year as an example, and breakdown the workload of the Court. In the 1982 October Term, there were 5,079 cases on the Court’s docket (Stumpf, p.397). 42 Therefore, the justices and their clerks (each justice has between two and four clerks) were responsible for wading through all of these cases. This sometimes requires doing additional research about the legal issues involved in a case and writing memos to other clerks or justices. The justices then vote at a Friday conference to determine which cases will be granted review. In 1982, this screen- ing process eliminated most of the cases, leaving 298 to be decided by the Court (p. 397). After a case has been accepted, both parties submit new briefs arguing the merits of their side, as well as reply briefs in response to their opponent’s argument (Baum, p.113). Other interested parties may also be permitted to submit amicus curiag briefs. The written material in the briefs is then often supplemented by oral arguments before the Court. After oral argument, the jus- tices generally cast a tentative vote on the case during a conference held later in the same week (p.114). A justice is then assigned to write the opinion of the Court. In 1982, of the 298 cases reviewed on the merits, 182 resulted in full written opinions and 116 cases were hand- led in summary fashion (brief pg;_ggr_i_g1n_ opinions),(Stumpf, p.398). Written opinions often involve negotiations between the justices, which affects the lan- guage of the majority opinion and who is included in the majority. Thus, the majority opinion may go through 43 several drafts and revisions. In addition, other justices may wish to write a concurring or dissenting opinion (Baum, p.116). The point of this discussion is to provide some evi- dence that the justices face a heavy workload. There are some scholars (e.g. Bickel, 1973; Casper and Posner, 1976), and even some justices (e.g. Warren and Douglas), who have felt that the Supreme Court is not overworked. However, it is not being argued that the justices are necessarily overburdened or overworked, but rather with approximately five thousand cases to deal with each year, they are kept quite busy. In fact, Casper and Posner’s study found that at most each justice can spend an average of 9.5 minutes per paid petition for certiorari and con- siderably less time on petitions filed in forma pauperis (1976, pp.65-66). Furthermore, this work is not usually of a trivial or mundane nature. It is often quite detailed and complex. How then do the justices deal with this arguably com- plicated and large amount of information when as Justice Burger has stated they have less "time and freshness of mind for private study and reflection...[and] fruitful interchange...indispensable to thoughtful, unhurried deci- sion"(O’Brien, p.153)? It would seem that the justices must, by necessity, develop some type of simplified deci- sion procedure. It is claimed in Bob Woodward and Scott 44 Armstrong’s book, The Brethren, that for Justice Brennan sorting through petitions to the Court was "like sepa- rating the weeds from the flowers in the garden"(1979, p.273). Brennan spent only ten to fifteen seconds on some petitions, because after sixteen years on the Court "he had developed a special feel for recognizing the important cases"(p.273). Chief Justice Warren (citing Justice Har- lan) has also asserted that whether a case is certworthy "is more a matter of ’feel’ than of precisely ascertainable rules"(Stumpf, p.399). Along similar lines, Justice Blackmun has admitted: the heavier the burden, the less is the pos- sibility of adequate performance and the greater is the probability of less-than-well considered adjudication...One, therefore, to a large degree, relies on innate and hope- fully already developed proper judicial reaction (Ripple, 1980b, p.175). The conclusion reached in this research, is that due to the amount of work before the Court and its complexity, the justices must often rely upon a simple decision making structure. They arguably do not have the time, resources, or intellectual capacity to make all of their decisions in a more comprehensive manner. The justices instead are often forced to behave in a boundedly rational or cognitive-cybernetic fashion. As discussed above, with a cognitive-cybernetic decision maker there is no attempt to be comprehensive or make extensive calculations. Instead, only a few critical variables or cues are focused on. The 45 individual relies upon previous experience and decision rules to aid in the decision making process. This more simplistic view of human decision making aptly explains Brennan and Warren’s "feel" and Blackmun’s "proper judi- cial reaction" in examining cases. This cybernetic—boundedly rational view also fits nicely into the Rohde-Spaeth framework of Supreme Court decision making. While this framework was written with rational choice theory in mind, it is quite compatible with the work of Simon and Steinbruner. As has already been mentioned, both the Rohde—Spaeth framework and Simon’s writings discuss the importance of attitudes and the particular facts involved. The attitudes and values held by a person determine what their goals are, and affect what facts are perceived. This combination of attitudes, goals, and facts to a great extent determine what decision is ultimately reached by the person (Rohde and Spaeth, of course, also consider the "rules of the game"). Steinbruner’s work further bolsters and fleshes out the importance of the particular situation facing the Court (or part three of the Rohde-Spaeth framework). The cognitive-cybernetic decision maker only focuses on a few incoming variables. These critical variables greatly affect how the individual responds in the situation. This limited focus results in a simplified decision making E... 46 process. Thus, the cognitive-cybernetic model and bounded rationality can be used to supplement Rohde and Spaeth’s work. Finally, this simplified approach to decision making can also provide a theoretical foundation for cue theory or fact models. It gives a cognitive basis for such explanations of Supreme Court behavior. Consequently, in many respects these different ideas and theories dovetail. Therefore, they have the potential to be unified and pro- vide a more complete view of Supreme Court decision making. In the following chapters of this dissertation, this combined approach will be specifically applied to the Establishment Clause cases. Chapter 3: A Theoretical Perspective on the Burger Court’s Establishment Clause Decisions "We are under a Constitution, but the Con- stitution is what the judges say it is..."-- Governor Charles Evan Hughes (later to be an associate and chief justice of the U.S. Supreme Court) A. Overview As was discussed in the previous chapter, the Rohde- Spaeth framework indicates that there are three main fac- tors which need to be considered when looking at Estab- lishment Clause cases. First, there are the individual justices involved in the decisions. As noted above, these individuals have freedom to act as policy makers and often do so. Therefore, their goals and policy preferences must be taken into account. The second factor, the institu- tional rules under which the justices work, at times, play a role in Supreme Court decisions. However, if a case has reached the stage of being formally decided by the Court, in almost all situations the justices will have complete freedom in deciding it as they see fit.[1] Since the research design being proposed here only considers for- mally decided cases, this factor does not play an impor- tant role in this work. The third factor, the facts of a case, is of great importance. There are particular facts or situations which seem to make a difference in how the Court rules. There are certain features which seem to greatly affect whether or not the Supreme Court upholds or strikes down a rule or 47 48 statute. Change the fact pattern and the decisions reached by justices may change as well. Thus, the factual situa— tion facing the Court should be assessed before one attempts to project the outcome of a case. Yet, while some legal scholars have noted the importance of a fact or facts to a particular case, none have specifically laid out all of the factors which can make a difference in Establishment Clause cases in general. Additionally, as should already be clear, this study views the justices as boundedly rational or cognitive— cybernetic decision makers. In almost every case there are countless facts or factors which could be considered. The justices do not have the time, resources, or intellectual capacity to consider every feature of every case. It will be argued that therefore the justices tend to pick out certain cues or facts which simplify the deci- sion they need to make. In doing this, the justices have created an internal formula or mechanism which aids them in their decision making process. Thus, when hearing a case on the Establishment Clause, certain facts will stand out as being important to a justice. This does not mean that what are considered to be important facts or factors cannot change, because the par- ticular values of a justice make a difference. Each jus- tice’s beliefs, attitudes, and biases are based on a few dozen values he or she holds (Spaeth, 1979, p. 129). These values affect what is viewed or perceived as being 49 important, and can act as a screening device when review- ing the facts of a case (as discussed earlier in connec- tion with the work of Herbert Simon). Some facts will be carefully considered, while others are quickly passed over. Consequently, if a justice’s values change, then the particular facts in a case he or she focuses on could also change. However, in most situations an individual’s values will remain relatively constant (Spaeth, pp. 120- 127). Likewise, John Steinbruner points out that it is "generally agreed that the basic structure of attitudes, once established, is very resistant to change . . ."(1974, p.102). This, of course, is closely tied to the cognitive principle "stability," which was discussed in the last chapter. So while the facts which are focused upon by any particular justice could change, it does not appear very likely. Upon accepting the idea that there are key facts or cues which guide the justices, how does this discussion then specifically relate to Establishment Clause cases? The main objective of this chapter is to offer a perspec- tive that will help to identify the factors which make a difference in these cases. This perspective will come primarily from the justices themselves and account for the varying situations which arise. Before turning to the more specific or narrower aspects of how the justices view these cases, it is important to consider the major themes 50 and concepts in this area of law. Let us begin with a brief, general overview of some of the broad issues involved here. This issue area revolves around the relationship between church and state. Viewed as a continuum, on one extreme there would be pp separation between church and state. Church and state would be closely tied in the affairs and decisions of one another. There would be an official state church or a certain set of beliefs would have the state’s endorsement or seal of approval. Either church or state could aid the other in any way desired. At the other extreme, there would be complete separation between the two. Here any connection, whatsoever, would be considered improper. Aid, no matter how minimal or indirect, would be disallowed. It is clear that the Establishment Clause bans the extreme of the continuum involving a state religion or no separation. This has been echoed repeatedly by the Supreme Court during the Burger years. Warren Burger wrote in the Lgmon v. Kurtzman decision that the "three main evils against which the Establishment Clause was intended to afford protection are: sponsorship, financial support, and active involvement of the sovereign in reli- gious activity"(1971, p.612).[2] The Court has also stated that this Clause attempts to prevent the intrusion of either church or state into the precincts of the other 51 (Lemon, 1971, p.614; Larkin v. Grendel’s Den, 1982,p.126; Lvnch v. Donnellv, 1984, p.672). It is at least a pledge that no single religion will be designated a state reli- gion for the union of church and state tends to destroy both, said the Court in School District of Grand Rapids v. Ball (1985, p.398). Additionally, Burger pointed out, in Lemon, that the authors of the First Amendment did not simply prohibit the establishment of a state church or a state religion, but instead commanded that there should be no law respecting an establishment of religion (1971, p.612). A law may be respecting the forbidden objective while falling short of its total realization, Burger explained, saying that a given law might not establish a state religion but nevertheless be one respecting that end in the sense of being a step that could lead to such establishment (1971, p.612). So, not only is the extreme position of a state religion inappropriate, but also laws or practices which can help lead to establishment, and which are closer to the center of the continuum. This further lessens the possible range of constitutionally acceptable practices. What about situations at the other end of the con- tinuum? The Court’s opinions have shown that this also is not where church-state relations can fall. Burger and the majority of the Court have asserted that the First Amend- ment does not say that in every and all respects church 52 and state shall be separated (Walz v. Tax Commission, 1970, p.669; see also gpmmittss for Public Education v. Nygpisp, 1973, p.760; Lynsh, 1984, p.672). They have stated that there cannot be absence of all contact due to the complexities of modern life, and in fact there are bound to be incidental benefits passed between them (flsls, 1970, p.676). The Chief Justice also later wrote in the Lsppp decision that total separation is not possible in an absolute sense. "Some relationship between government and religious organizations is inevitable" (1971, p.614). This then brings us to what is permissible between church and state. The Court calls it "benevolent" neutrality. As should be clear this falls somewhere between laws respecting an establishment of religion and complete separation. The Court claimed in Gillette v. ups; that perhaps the central purpose of the Establishment Clause is to insure government neutrality in matters of religion (1971, p.449). In gals, the majority stated that the government should be involved in neither the sponsor- ship of religion nor hostility toward it (1970, p.672). Similarly, government power should not be used to handicap religions or favor them (McDaniel v. Patv, 1978, p.639). However, government neutrality in this area of law is not as straightforward as it appears at first blush. The Supreme Court has held that neutrality "is not so narrow a channel that the slightest deviation from an absolutely 53 straight course leads to condemnation"(Walz, 1970, p.669). Burger writes that there is "room for play" with this legal concept and calls it "benevolent" neutrality (p.669). In other rulings, the Court has espoused that "neutrality in matters of religion is not inconsistent with benevolence" and the Constitution "mandates accom— modation, not merely tolerance of all religions, and for- bids hostility toward any" (Gillette, 1971, p.454; Lypsp, 1984, p.673). What is seen in these decisions is that while neutrality is a guiding principle it is not absolutely strict or blind to religion. There is a zone of allowed accommodation. The neutrality which is required need not stem from a callous indifference, but may at times be benevolent. In fact, if it was absolutely strict or indifferent in all cases, this could be viewed as being hostile toward religion. This leads directly to a second part of this concept of neutrality; incidental benefits which are accorded to persons or institutions with some religious connection. As stated above, the Court does not believe it is possible to have complete separation of church and state. For example, to have complete separation, the government could not provide a church or religiously affiliated school with fire or police protection. Such aid has not been viewed as sponsoring or endorsing religion, and in fact, if not 54 provided, the government could be seen as being hostile toward religion. Thus, benevolent neutrality allows for "neutral nonideological aid" which only "indirectly and incidentally" promotes a religious function (Nygpisp, 1973, p.775). The discussion up to this point has been presented in order to provide a general overview of what the broad questions and issues facing the Supreme Court are. Yet, knowing the Court finds certain neutral practices con- stitutional is only a starting point. For while relying on this general principle, the Burger Court has also laid down a three-part test for Establishment Clause cases.[3] First, a statute must have a secular legislative purpose; second, its principal or primary effect must be one that neither advances nor inhibits religion; third, it must not foster an excessive government entanglement with religion (Lemon, 1971, p.612). Additionally (although this may be part of the entanglement test) the statute must not create an excessive degree of political division along religious lines (p.622). This three-part test attempts to touch upon the key concepts involved in this area, including neutrality. However, the purpose-effect-entanglement test also does not put this matter to rest. For in Lsmpp Burger admits that "we can only dimly perceive the lines of demarcation in this extraordinarily sensitive area of con- 55 stitutional law,"(1971, p.612). He goes on to say that "the line of separation, far from being a ’wall,’ is a blurred, indistinct, and variable barrier depending on all the circumstances of a particular relationship," (p.614). In later cases, the Court says that this three-prong test provides "no more than a helpful signpost" in dealing with Establishment Clause cases (fippp v. McNair, 1973, p.741; Mueller v. Allen, 1983, p.394). It "serves only as guidelines" to the necessary constitu- tional inquiry (Meek v. Pittenqer, 1975, p.359; School District of Grand Rapids, 1985, p.383; see also Tilton v. Richardson, 1971, p.677; Lynch, 1984, p.678). These statements underscore two very important points. First of all, while the purpose-effect- entanglement test has value, to some extent, it (like the benevolent neutrality principle) is vague. It lacks specific or precise detail, and therefore it can be no more than a "helpful signpost." Secondly, the justices do consider the particular factual situation before them. Their decisions depend upon "all the circumstances of a particular relationship." This can be tied to cybernetic decision making. The claim being that the justices rely upon cues or specific facts before arriving at a decision. While general principles of law or broad tests may be of some help, often a justice will focus in on certain fac- tual information. 56 B. Establishment Clause Typology Since the Supreme Court itself admits that the benevolent neutrality doctrine and the purpose-effect- entanglement test do not fully capture how the Court reaches its decisions, this research proposes the use of a more specific typology or classification scheme in an attempt to explain these cases. This typology comes from the Court’s own written opinions. Its major headings are derived from a three-part test given in Lemon, yet dif- ferent from the purpose-effect-entanglement test. In addition, the various components of the typology are based upon and bolstered by statements made in numerous Estab- lishment Clause cases.[4] Common themes run throughout this area of law, and the concepts used here were repeatedly supported. In the Lemon decision, Chief Justice Burger writes: In order to determine whether the government entanglement with religion is excessive, we must examine the character and purposes of the institutions that are benefited, the nature of the aid that the State provides, and the resulting relationship between the government and the religious authority (1971, p.615). From this "excessive entanglement" test three basic issues or types of questions are derived. First, what is the specific aid or practice being proposed? More particu- 57 larly, what is its nature and purpose? Second, who is getting the aid or will be affected by this practice? In other words, what is the character, purpose, and history of the institutions which will benefit from the aid or activity? Third, what is the resulting relationship between the government and religion? Precisely how much and what type of contact will there be between church and state if this aid is provided or activity allowed? It will be argued here that these three basic ques- tions or areas cover not only excessive entanglement, but the entire inquiry into church-state relations. These questions raise the key and fundamental issues confronting the Court. They provide the essential factual information necessary for deciding these cases. It should be noted that the major concepts of the purpose-effect-entanglement test are incorporated into this typology, but in a more detailed and straightforward manner. Consequently, benevolent neutrality is also covered by it. Lastly, it will also be argued that in addition to these core Estab- lishment Clause questions, one may want to consider other complicating issues. Some literature indicates that there may be cues or factors beyond the bounds of the Estab- lishment Clause which could potentially influence the out- come of a case in this area. With this introduction, each topic will now be delved into more deeply. 58 1. What is the Aid/Practice (the Nature of the AidZPractice) ? As the previous discussion indicates, three types of factual questions or areas need to be studied. As part of this, specific aspects of each will be assessed in detail and some testable propositions put forth. The first of these factual areas concerns the nature of the aid or practice. As should be expected, the members of the Court have an interest in precisely what is being proposed. What is being given? Why is it being given? What is its purpose? Is the nature of the aid one of general state interest or does it have religious motivations? In Roemer v. Board of Public Works, the Court writes that the state must confine itself to secular objectives (1976, p.747). The Court in Lypsp proClaims that legisla- tion or governmental action can be struck down when a secular purpose is lacking (1984, p.680). It asks the question of whether or not the aid or activity is "a pur- poseful or surreptitious effort to express some kind of subtle governmental advocacy of a particular religious message"(p.680)? The majority says in Gillette that the Establishment Clause prohibits the government from depart- ing from secular purposes in order to put an imprimatur on one religion or religion as such (1971, p.450) . In Ks;- Laps v. Jaffree, the Court states that it is appropriate to ask whether the government’s actual purpose is to 59 endorse or disapprove of religion. A statute which is motivated in part by religious purpose may satisfy the Establishment Clause--but if it is entirely motivated by the purpose to advance religion, it must be invalidated (1985, p.56). Thus, the Supreme Court has made clear in a number of cases that if a statute or practice has an essentially religious purpose, it is unconstitutional. For example, the Burger Court struck down statutes in Wallace and sppps v. Graham, where they found there was no secular purpose, but instead the laws were motivated by the purpose to advance religion (1985; 1980). Accordingly, it is hypothesized that where the sole (or predominant) purpose of a law is religious, the Court will find such a law to be unconstitutional. As was just shown, when considering the particular aid or practice before it, the Court examines the purpose of the activity. There is also a second aspect to the Court’s assessment. This relates to the nature of the aid itself, and is directly tied to the idea of incidental benefits, which was previously discussed. As mentioned above, the Supreme Court has espoused the idea that while the Establishment Clause was meant to prevent the intrusion of church and state into the precincts of the other, total separation is not possible. "Not every law that confers an indirect, remote, or 6O incidental benefit upon religion is constitutionally invalid"(Lyps_h, 1984, p.683; Meek v. Pittenqer, 1975, p.359; see also Widmar v. Vincent, 1981, p.273). Using the Court has accepted "neutral, non- similar language , ideological aid, which only indirectly and incidentally promotes a religious function"(Nygpist, 1973, p.775) . Following this principle, Walter found certain general health services provided to non-public schools to be constitutional (1977). The majority’s rational was that these services have no educa- and therefore do not create an tional content, They impermissible risk of fostering ideological views. were simply the provision of a general welfare service to the community. Along similar lines the Court has held that policemen protecting children from traffic hazards or and "fire inspections, building and zoning regulations, state requirements under compulsory school-attendance laws are examples of necessary and permissible contacts"(Walz, 1970, p.671; Lemon, 1971, p.614). Thus, when considering the nature of the aid, the Supreme Court examines how indirect, remote, and inciden- tal the benefits it provides to religion are. To be acceptable, it must be of a nature which simply aids the general health, welfare, or needs of citizens. Accor- dingly, it is hypothesized that when aid fits the descrip- 61 tion of being a nonideological, general, welfare service, the Court will find it to be constitutional. 2. Who is Receiving the Aid or is Involved with the Practice? The second main factual question which needs to be assessed, concerns who receives the aid. The Court in School District of Grand Rapids writes: "Our inquiry must begin with a consideration of the nature of the institu- tions in which the programs operate"(1985, p.384). As would be expected, the Supreme Court in not only inter- ested in the aid itself but also the parties getting aid or affected by it. This information is essential in deciding if the aid or practice is proper or not. What is the character of the institutions receiving aid or involved with the practice? Is aid being given to institutions in a neutral fashion? Is the aid directed at minors or adults? Is there any prior history or tradition indicating whether a group or institution should be pro- vided with aid? As mentioned above (while introducing the concept of benevolent neutrality), the Court has claimed that the central purpose of the Establishment Clause is to insure government neutrality in the area of religion (Gillette, 1971, p.449) . The Court has echoed this theme on numerous occasions. First of all, no particular religion should be 62 favored or preferred, and none interfered with (Walz, 1970, p.677; Larson v. Valente, 1982, p.244; Harris v. McRae, 1980, p.319) . The government may not use religion as a basis for the imposition of duties, penalties, privileges, or benefits (McDaniel v. Patv, 1978, p.639). Furthermore, while there is to be no favoritism nor discrimination among religious sects, there is also to be none between religion and nonreligion (Walz, 1970, p.695; School District of Grand Rapids, 1985, p.381). The government needs to be evenhanded in its treatment of those with religious beliefs and those without. In Welsh v. U.S. , in a concurring opinion, Justice Harlan asserts that Congress "cannot draw a line between theistic or non- theistic religious beliefs on the one hand and secular beliefs on the other," since any such distinctions "are not compatible with the Establishment Clause"(1970, p.356). In keeping with the principle of neutrality, the Cmnt.has repeatedly considered whether the aid being examined benefits a broad class of individuals (Walz, 1970, p.673; Welsh, 1970; Mueller v. Allen, 1983). Jus- tice Harlan has written: "the critical question is whether the scope of the legislation encircles a class so broad that it can be fairly concluded that all groups that could be thought to fall within the natural perimeter are included" (Wslsp, 1970, p.357; see also Wsls, 1970, osi - _: 63 p.717) . More specifically, in order to be thought of as being sufficiently broad, the legislation assisting reli- gious groups must accord nonreligious groups the same benefits (flslsh, 1970, pp. 356-361). The program or aid should be open to all (Widmar, 1981; Witters v. Washington Department of Services, 1986) . Applying this to the area of aid to schools, the Court has stated the importance that the class of beneficiaries include all schoolchildren (Le_m_o__n_, 1971, p.616; Nygpist, 1973, p.775; M_ee_k, 1975, p.362). It mat— ters whether the aid is given to both public and nonpublic schools (Lsppp, 1971, p.643; Tilton, 1971; Aguilar y; Felton, 1985). Moreover, in Sloan v. Lemon, it is men- tioned that there could be problems when benefits from a program do not aid all parents (1973, p.832). Consequently, it should be clear that the Supreme Court takes the concept of neutrality quite seriously. In Establishment Clause cases treating all groups even- handedly can play a pivotal role in the outcome of cases. Therefore, it is hypothesized that when legislation in this area aids or affects all groups equally, it has a far better chance of being upheld by the Supreme Court. It has already been mentioned, however, that the Supreme Court does not always follow the policy of absolute neutrality. There are time when the Court allows governmental practices which are benevolent toward reli- 64 gion. Accommodation is espoused rather than strict neutrality in certain situations. This brings us to the second aspect concerning the parties involved. The issue here is whether there is a long history of the government providing aid to or allowing certain practices within par- ticular institutions. The importance of history and tradition was made quite evident in the Burger Court’s first Establishment Clause case. Burger ,writing for the Court, cites Justice Holmes: "a page of history is worth a volume of logic"(flsls, 1970, p.676; also quoted in Nyguist, 1973, p.777). Burger goes on to say that while no one acquires a protected right in violation of the Constitution by long use, "an unbroken practice . . . is not something to be lightly cast aside"(Wsls, 1970, p.678; see also Marsh y; Chambers, 1983, p.790). The more long-standing and widely accepted a practice, the greater its impact upon constitu- tional interpretation (Wsls, 1970, p.681). More specifi- cally, if something has been practiced for two hundred years, by common consent, it will require a strong case to affect it (p. 678). Such long term practices can become "deeply embedded in the fabric of our national life . . ."(p. 676; see also Mspsh, 1983, p.792). There are other reasons for allowing aid or practices which are historical. One reason is that if the practice or aid dates to the early years of this nation’s existence Fe; :1 65 it possibly reflects "the understanding of our Founding Fathers"(Wsls, 1970, p.680). Justice Brennan states in flsls that this is "a fact of considerable import"(p.681). This notion has been restated in later cases. In Mspsp, the Court claims that historical patterns shed light on what the draftsmen of the Constitution intended the Estab- lishment Clause to mean (1983, p.790). Acts of the First Congress (many of whose members had taken part in the framing of the Constitution) should be treated as "con- temporaneous and weighty evidence of the Constitution’s true meaning"(p.790). Likewise, the Court in Lypsp says that constitutional decisions of the First Congress "are of the greatest weight" in the interpretation of the Con- stitution (1984, p.674). Another reason for allowing these historical activities to continue is simply because they have a track record. The Court knows what to expect. The particular institutions involved have shown over the years how they will respond. The Court knows that these participants and activities have not led to the establishment of a religion in this country. For example, the issue before the Court in Wsls concerned the constitutionality of tax exemptions for churches (a practice going back to the time of the Framers). In the majority opinion for the case, Burger writes that if this is the first step toward the estab- 66 lishment of religion "the second step has been long in coming"(1970, p.679). This leads to the conclusion that the justices con- sider the history of the institutions and activities which come before them in Establishment Clause cases. They reflect on how deeply ingrained certain types of aid and practices are. In some cases the Court will accommodate these long time practices and thus act in a benevolent fashion. Therefore, it is hypothesized that if a certain act or practice has a long history or tradition, the Court is far more likely to find it acceptable constitutionally. One final aspect of this second part of the typology needs to be examined. This deals with the age and maturity of those affected by the aid or included in the activity. In addition, it covers the purpose and motiva- tions of the institution which is directly responsible for distributing the aid. These questions arise almost exclu- sively in the area of parochiaid. In discussing nonpublic elementary and secondary schools in Wsls, the Court declares these schools "plainly tend to assume future adherents to a particular faith by having control of their total education at an early age" (1970, p.671). In Lsppp, Burger says church-related elementary and secondary schools have a "religious mis— sion" and are dedicated to rearing children in a specific faith (1971, pp. 613 and 618). The Court ,in Aguilar v. 67 Felton, refers to the "pervasively sectarian environment" of these schools (1985, p.412). Likewise, Justice Douglas asserts in Lsmpp that it "is well known that everything taught in most parochial schools is taught with the ultimate goal of religious education in mind"(1971, p.634). In parochial schools religion permeates the whole curriculum (p.634; see also Qpppittee for Public Education yp_3sgsp, 1980). These schools give churches the opportunity to indoctrinate children with their creed (Lsppp, 1971, p.631). He concludes that the "psispp d’etre of parochial schools is the propagation of a reli- gious faith"(p.628). Almost diametrically opposed to this view of primary and secondary schools is the Supreme Court’s perception of nonpublic colleges and universities. According to the Court, church-related colleges perform essentially secular educational functions (Roemer v. Board of Public Works, 1976). Religious teaching is not pervasively intermixed with each and every activity. It does not so permeate the curriculum as to make their secular and religious func- tions inseparable (Tilton, 1971, pp.680-681, Roemer, 1976, p.759). In Tilton, Burger (writing for the Court) claims that religious indoctrination is "not a substantial pur- pose or activity" of these institutions of higher learning (1971, p.687). In fact, these institutions exist in an "atmosphere of academic freedom rather than religious 68 indoctrination"(p.682; compare this to Burger’s discussion of primary-secondary school teachers in Lsppp, 1971, p.618). Lastly, college students are less impressionable and less susceptible to religious indoctrination than are students in primary and secondary schools (Tilton, 1971, p.686; Grand Rapids School District, 1985, p.383). In addition, the Court has mentioned situations involving adults but outside of an academic setting. In the case of Marsh v. Chambers, the Court alleges that since the individual claiming injury by the practice is an adult, he "presumably is not readily susceptible to reli- gious indoctrination or peer pressure"(1983, p.792). The individual in the case could absent himself without incur- ring any penalty. Similarly, Justice O’Connor in'a con- curring opinion in Wallsss v. Jaffree, expresses that it matters whether a practice is "primarily directed at adults, who presumably are not readily susceptible to unwilling religious indoctrination"(l985, p.81). She goes on to say that "this Court’s decisions have recognized a distinction when government-sponsored religious exercises are directed at impressionable children who are required to attend school, for then government endorsement is much more likely to result in coerced religious beliefs"(p.81). Consequently, it appears that the justices examine the age and how impressionable the individuals involved in the case might be. As part of this, the particular 69 institution involved and its motivations are considered. Accordingly, it is hypothesized that the Supreme Court is far more likely to accept practices involving colleges and universities or directed at adults. 3. What is the Resulting Relationship? The third area of the typology can now be considered. What is the resulting relationship between government and religion? Upon discussing what the aid is and who is get- ting it, the next logical question is: what type of con- tact will there be between church and state if this action is allowed? This directly and explicitly strikes at the heart of the separation of church and state issue. Two aspects of this question will be looked at (both of which relate directly to the third part of the purpose-effect- entanglement test). In a number of cases, the Supreme Court has made clear that if the involvement between church and state is too close or intimate, it is improper. For example, the justices have concluded that it is a problem when government intrudes into the "everyday affairs" or "reli- gious affairs" of church-related institutions (flppp_yp McNair, 1973, pp.753-754; Alamo Foundation v. Secretarv of Labor, 1985, p.306). With this in mind, the Court laid down the "no excessive entanglement" requirement of the three-prong Establishment Clause test (Walz, 1970, p.664). ‘3 70 Along with this, the justices began raising questions in a somewhat new direction. In Wsls, Chief Justice Burger asks whether government involvement, in the activity under scrutiny, is excessive (1970, p.675). Is the involvement "a continuing one call- ing for official and continuing surveillance"(p.675)? Does it require "a sustained and detailed administrative relationship"(p.675)? In a concurring opinion, Justice Brennan mentions the undesirability of having extensive state investigations into church operations and finances (p.691). Following this lead, similar inquiries were made in subsequent decisions. In Lsppp, the justices rejected aid which involved strict government controls and surveillance by state authorities. Burger writes that the aid package creates "an intimate and continuing relationship" between church and state, since the state would be allowed to "inspect and evaluate a church-related school’s financial records . . ."(1971, p.621). The Chief Justice expresses that this kind of state inspection is fraught with Estab- lishment Clause difficulties. "A comprehensive, dis- criminating, and continuing state surveillance" is the sort of entanglement the Constitution forbids (p.620; see also Mssk, 1975; Aguilar, 1985; Thorton v. Caldor, 1985). Likewise, Justice Douglas ,in a concurring opinion, claims that this aid would put a "public investigator into every I: 7 _.—_ :5 71 classroom and entails a pervasive monitoring of these church agencies by secular authorities"(Lsmpp, 1971, p.627). Such "policing" of sectarian schools results in "vast governmental suppression, surveillance, or meddling in church affairs"(p.634). The Court has repeatedly pointed out the potential problems of government audits or on-site inspections of church-connected institutions (e.g. Tilpgp, 1971; apps, 1973; Committee for Public Education v. Regan, 1980). The Court is wary of situations where the state is required "to monitor the day-to-day activities" or "engage in onerous, direct oversight" of religious institutions (gsgsp, 1980, p.659). This type of "ongoing public inspection" leads to "a permanent and pervasive state presence"(Aguilar, 1985, pp. 412-413). Along these same lines, the Court has warned of the difference between personnel (usually teachers) and books. Burger, in the Lsmpp decision, asserts that teachers have a substantially different ideological character from books. "In terms of potential for involving some aspect of faith or morals in secular subjects, a textbook’s content is ascertainable, but a teacher’s handling of a subject is not"(1971, p.617; see also Wolman v. Walter, 1977; Bsggp, 1980; National Labor Relations Board v. Catholic Bishop of Chicago, 1979). Burger also writes that "unlike a book, a teacher cannot be inspected once so as to determine the 72 extent and intent of his or her personal beliefs and sub- jective acceptance of the limitations imposed by the First Amendment"(Lsmgp, 1971, p.619). The concern being that teachers may intentionally or inadvertently become involved in inculcating particular religious beliefs (e.g. stitt v. Committee for Public Education, 1973, p.480; ngmgp, 1977, p.247; sgpggl_pistrict of Grand Rapids, 1985, pp.385-386; Mggk, 1975, p.370). Justice Douglas alleges that "we deal not with evil teachers but with zealous ones who may use any opportunity to indoctrinate a class" (Lsmpp, 1971, p.635). For this reason, when governmental aid involves teachers, there is a need for surveillance or intimate contact to ensure that what is being taught is permissible. Yet, such contact also runs afoul of Establishment Clause restrictions (Lspgp, 1971, p.619). Thus, those wanting to provide such aid are faced with a "no win" situation. The preceding words of the justices indicate that they consider the resulting relationship between church and state when aid is proposed. If the relationship is too close or intimate there can be problems. If the state must act as a watchdog, involvement into the affairs of church-related institutions has become far too intrusive. Therefore, it is hypothesized that the Supreme Court will strike down any law or practice requiring governmental 73 surveillance or significant inspection of religious institutions and their financial records. A second aspect of the resulting relationship between church and state has already been mentioned in a number of the above quotations. This concerns whether the government’s involvement continues into the future or not. For example, in Lspgp, the Court mentions the "intimate and continuing relationship" and "continuing state sur— veillance"(l971, pp.621 and 620). Likewise, teachers need to be watched on a continuing basis while a book can be inspected only once (p.619). In ngs, Burger states: "the questions are whether the involvement is excessive, and whether it is a continuing one calling for official and continuing surveillance . . ."(1970, p.675). Directly opposed to aid given in a continuing or continuous fashion, are benefits given only once. For example, in Tilton v. Richardson and Hunt v. McNair, the aid provided was a "one-time" grant with "no continuing financial relationship" between government and church- related institutions (1971, p.688; 1973, p.754). In both cases, this was stated as one of the reasons for upholding the governmental practice. Why should it matter whether aid is provided only once or may continue to be provided in the future? The Court, on numerous occasions, has given its reasoning for such a distinction. The justices mention, in the Lemon pit: 74 decision, that they are confronted with an aid package which will lead to "successive and very likely permanent annual appropriations that benefit relatively few reli- gious groups"(1971, p.623). Thus, political divisiveness and fragmentation are likely to be intensified along reli- gious lines (p.623; ngs, 1970, pp.695-698). In addition, the potential for political divisiveness tied to religious belief is aggravated by the "need for continuing annual appropriations and the likelihood of larger and larger demands as costs and populations grow"(Lemon, 1971, p.623; Sleep, 1973. p.831; Exquist. 1973, p.796; Meek, 1975,p.365). As stated earlier, the Court in Lsmgp added to the purpose-effect-entanglement test the requirement that a statute must not create an excessive degree of political division along religious lines (1971, p.622). The Court has never specifically stated whether this is a fourth factor to the test or merely an extension of the third prong. Yet it is clear that the justices assess this fac- tor. They claim some types of aid result in competition and strife among religious sects and can ultimately bring a political system to the breaking point (Nygpist, 1973, p.796; see also ngs, 1970, p.694). In Lspgp, the Court declares: Ordinary political debate and division, however vigorous or even partisan, are normal and healthy manifestations of our democratic system of government, but politi- 75 cal division along religious lines was one of the principal evils against which the First Amendment was intended to protect (1971, p.622). Thus, the justices have expressed concern over frac- tionalizing the electorate and officeholders by religious belief and practice. Such division is seen as a threat to the normal political process. Further, it is alleged that aid involving annual or continuing appropriations is more likely to lead to such political divisiveness. Con- sequently, it is hypothesized that if aid is given as a one-time grant rather than on a continuing basis to reli— gious organizations, it is far more likely to be upheld by the Supreme Court. This completes the review of the Establishment Clause issue Typology. In summary, it consists of three primary questions: (1) what is the aid or practice being con- sidered? (2) who is involved or receiving the aid? (3) what kind of relationship between church and state does this result in. Further, each question or area can be broken down into smaller elements. However, in addition to this typology, one should consider a few final factors before attempting to explain and predict these decisions. There are factors or issues which fall somewhat beyond the bounds of the Establishment Clause per se, but which may potentially influence the outcome of a case in this area. Fri- 4 . 76 C. Potential Complicating Issues William Cohen and John Kaplan write that the Estab- lishment and Free Exercise Clauses of the First Amendment were not designed to serve contradictory purposes. "They have a single goal--to promote freedom of individual reli- gious beliefs and practices"(Cohen and Kaplan, 1982, p.411). Despite this, these scholars claim there is "an uneasy tension" between the two clauses (p.411). They are not alone in this assessment. Others have also discussed the "tension," "serious tension," or "natural antagonism" between the First Amendment’s two religion clauses (e.g. Choper, 1986, p.1657; Tribe, 1978, p.815; Nowak, Rotunda, and Young, 1978, p.849). For example, consider a situa- tion where a religious group desires access to a city park in order to hold a religious meeting. Does the Free Exercise Clause require a city to allow such a meeting or does the Establishment Clause prohibit such use of public property (Cohen and Kaplan, 1982, p.411)? Subsequently, Laurence Tribe concludes that the religion clauses "which for the framers represented relatively clear statements of highly compatible goals, have taken on new and varied meanings that frequently appear to conflict" (1978, p.812). The justices on the Supreme Court have also, at times, acknowledged this tension between the clauses. In Walz, Burger states: 77 The Court has struggled to find a neutral course between the two Religion Clauses, both of which are cast in absolute terms, and either of which, if expanded to a logi- cal extreme, would tend to clash with the other (1970, p.668). Justice Harlan, writing in concurrence, insists that the function of the Establishment Clause is not wholly auxiliary to the Free Exercise Clause since "it bans some involvements of the State with religion that otherwise might be consistent with the Free Exercise Clause"(Welsh yp_gp§p, 1970, p.374). The Court in Widmar v. Vincent asserts that the state’s interest in achieving the sepa- ration of church and state "is limited by the Free Exercise Clause"(1981, p.276). While the Court has mentioned the potential clash between these clauses, it has never specifically indicated the dominance or preferred position of one clause over the other. However, Laurence Tribe claims that the free exercise principle should be dominant in any conflict with the anti-establishment principle. "Such dominance is the natural result of tolerating religion as broadly as pos- sible rather than thwarting at all costs even the faintest appearance of establishment"(1978, p.833). He believes that when the Free Exercise and Establishment Clauses con- flict "support of the former would be more faithful to the consensus present at the time of the Constitutional Con- vention and of the First Congress"(p.819). In reviewing E, tun-“fl 78 the cases in this area, Tribe concludes that when actions are "arguably compelled" by free exercise they are not forbidden by the Establishment Clause (p.822). The Supreme Court has allowed the Free Exercise Clause to carve out of the area of establishment, a zone of permis- sible accommodation (p.823). Similarly, Jesse Choper also sees situations where free exercise is preferred. Accord- ing to Choper the Court has sometimes held that "the free exercise clause ppligss government to act with a non- secular purpose--actually, to give a preference to religion--when the action is necessary to permit the unburdened exercise of religion"(1986, p.1652). Therefore, while free exercise, without further examination, would appear to be a separate issue from establishment, this is not always the case. These issues may conflict, at times, and this could influence the out- come of a decision. There is some research indicating that the Court ,in certain situations, favors free exercise over establishment. Accordingly, it is hypothesized that the Court would be somewhat more likely to decide a case in favor of a litigant stating a free exercise claim. Another possible complicating influence in Estab- lishment Clause decisions concerns the involvement of the U.S. Government in a case. The bulk of previous research indicates that the federal government seems to enjoy an 79 advantage before the Supreme Court (e.g. Tanenhaus 1960 and 1963; Werdegar, 1967; Scigliano, 1971; Cannon and Giles, 1972; Carrington, 1974; Puro, 1981; O’Connor, 1983; Segal, 1984 and 1988). For example, from 1979 to 1983, the Court usually decided in favor of the federal government’s position where it appeared as a litigant or as an amicus curiae. Depending on the year, the government’s winning percentage ranged from 66 to 83 per- cent (U.S. Department of Justice, 1985). Likewise, Jef- frey Legal, examining only amicus curiae situations, found the solicitor general to consistently win an overwhelming majority of his cases (1988, p.138). He assessed solicitor general success from Eisenhower to Reagan. He found that no administration won less than 65 percent of its cases as an amicus, and the Kennedy administration had won a high of 87.5 percent of its cases (p.138). There are a number of possible explanations for this high success rate. It may simply be the result of the special relationship which exists between the Court and the Attorney General’s Office. The United States is the most frequent litigant which appears before the Court. Mark Galanter has found that "repeat players" have certain advantages and tend to win more than "one shotters" do (1974). In addition there is the solicitor general’s con— siderable and often highly rated staff (Legal, 1988, p.138). Justice Brennan has stated that the "ablest advo- 80 cates in the U.S. are advocates in the Solicitor General’s Office"(O’Connor and Epstein, 1983). Subsequently, Karen O’Connor claims that the solicitor general has "extraordinary influence" and his amicus briefs "have a substantial effect on public policy"(1983, p.264). Other commentators have dubbed the solicitor general as the Court’s "nine and a half" member or the "tenth jus- tice"(Werdegar, 1967; Scigliano, 1971; Ulmer and Willison, 1983; Caplan, 1987). Other possible explanations concern how the Supreme Court sees its role in our system of government. The Court may view itself as being partners with Congress and the President at the federal level. In Grit v. Wolman, for example, a federal district court stated that the fed- eral courts should have a "deferential attitude" toward Congress generated by "respect for a co-equal branch of government"(1972, p.744). Lastly, the Court may be attempting to stress the supremacy of the federal government over state and local government. Some litera- ture on the Supreme Court has pointed in this direction (Kaplan, 1972; Abraham, 1980; Henschen, 1983; Wasby, 1984). In conclusion, it appears to matter whether the fed- eral government is participating in a case or not. It may be the quality of work done by the solicitor general, or some broader reason. Yet in any case, the U.S. as a party 81 to a case or an amicus, tends to do better than other litigants. Therefore, if the U.S. is involved in an Establishment Clause decision, it is hypothesized that the Supreme Court is more likely to decide the case in favor of the position argued by the federal government. All of the key factual elements or cues which are to be considered in this work have now been discussed. In the next chapter, a specific Establishment Clause model will be proposed and operationalized. It will be derived from the ideas and concepts written of in this chapter. Additionally, each of the individual justices will be part of this model. This research is based upon the notion that it matters who is on the Court and how they perceive and react to the information given them. Thus, their specific biases and predispositions must be taken into account. Chapter 4: Operationalization and Methodology "More generally, if the data permit the use of more refined methods, certainly they should be employed. In any case, no method can predict Supreme Court decisions per- fectly."—-Fred Kort In the last chapter, nine hypotheses were stated relating to the Establishment Clause cases decided during Warren Burger’s tenure. In this chapter, those nine hypotheses will be operationalized into independent vari- ables which can be measured and tested. In addition, the dependent variable and twelve other "Justice" independent variables will be discussed and operationalized. These variables will then be used to construct an Establishment Clause model. The data and methodology to be used to test this model will also be identified. A. OperationalizationZMeasurement, Model, and Data Before discussing the independent variables to be operationalized for an Establishment Clause model, first the dependent variable will be defined. The primary dependent variable1 for this research is the vote of the individual justices in each Supreme Court decision involv- ing the Establishment Clause. The basic question before a justice is whether the statute or practice involved in the case violates this part of the Constitution. Therefore, 82 83 each decision can be seen as having one of two possible outcomes: a violation of the Establishment Clause or no violation. The dependent variable, VOTE, was coded in a com- patible dichotomous manner. It was coded as a "1" when the justice took an "accommodationist" stance, and "0" otherwise. In a vast majority of Establishment Clause cases, the government is attempting to aid or accommodate religion in some fashion, and the question is whether the government has gone too far. Has it breached the wall of separation by giving too much aid or allowing too much contact between church and state? In these cases, VOTE was coded "1" when the justice voted to uphold the law or practice. In a small percentage of the cases, the government has placed special restrictions on religion. It has erected a "high wall" in an attempt to guarantee sepa- ration. Instead of being accommodationist, this might be viewed as being "separationist". For example, in Widmar v. Vincent (1981), a state university refused to grant a student religion group access to its facilities while allowing all other organized groups such access. In such a case, VOTE was coded "1" if the justice voted to strike down the law or practice. The nine hypotheses which were elaborated upon in the last chapter will now be operationalized. The first of those nine hypotheses stated that where the sole (or 84 predominant) purpose of a law is religious, the Supreme Court will find such a law to be unconstitutional. The Court has declared that if a statute or practice has an essentially religious purpose, it will be struck down. The independent variable drawn from this hypothesis will be named PURPOSE. This variable is the most difficult variable to operationalize of those in this research pro- ject. This is due to the degree of judgment which is required of the researcher. There is no absolutely objec- tive way to measure the purpose of a piece of legislation or a public policy. At some point, the researcher must make a judgment call. Aware of this problem, an attempt was made to mini- mize the subjective nature of this variable as much as possible. It should be noted that the Court’s purpose- effect-entanglement test requires that a statute must have a secular legislative purpose in order to be constitu- tional. Arguably, it is always possible to find some secular purpose for any piece of legislation. Instead, the focus here will be on whether there is a religious purpose for the statute or practice. Furthermore, as stated in the hypothesis, this must be the sole or predominant reason for this activity. While not com- pletely objective, this does create a high barrier to be overcome before claiming that such a factual situation exists. 85 PURPOSE was coded in a 1/0 fashion. If the purpose of the law was viewed as clearly or predominantly attempt- ing to advance religion it was coded "1", otherwise it was coded as "0". It thus required an extreme and uncommon factual situation for this variable to be coded as a "1". For example, in Lemon v. Kurtzman (1971), where parochial schools were to be given funding for textbooks, instruc- tional materials, and teacher salaries, this variable was coded "0". Whereas, in a case where a law required a moment of silence for prayer or meditation in public schools (Wallace v. Jaffree, 1985), or the posting of the Ten Commandments (Stone v. Graham, 1980), PURPOSE was coded "1". The second hypothesis from the last chapter stated that when aid fits the description of being a nonideologi- cal, general, welfare service, the Court will find it to be constitutional. A practice should be deemed acceptable if it is this type of general government service. The second independent variable, GENERAL GOVERNMENT SERVICE, was drawn from this hypothesis. It was operational- ized in the following manner. Eight permissible types of aid were derived which could fit under this broad heading. The eight types of aid are: fire protection, police pro- tection, reimbursement for student transportation to and from school, the loaning of secular textbooks to students, school breakfasts and lunches, diagnostic health services, ‘~ ‘4’ 86 standardized state tests, and costs associated with taking school attendance. The fire protection, police pro- tection, student transportation, and loaning of textbooks can all be considered general services that the government provides to all of its citizens regardless of religious beliefs. They are community-wide services. The meals and diagnostic health services are general health and welfare services of the state. The bus transportation can also be seen as a health and welfare service which helps safeguard students from traffic injuries. The last two types of aid are included because they are general requirements of the state, and thus, it is not unreasonable to expect the state to have to pay for them. When any of these factual situations arose, GENERAL GOVERNMENT SERVICE was coded "1", otherwise it was coded as "0". The third independent variable is NEUTRAL. This comes from the third hypothesis which stated that when legislation in this area aids or affects all groups equally, it has a far better chance of being upheld by the Supreme Court. This hypothesis, of course, is largely based on the concept of benevolent neutrality. NEUTRAL was also coded as a 1/0 variable. It was coded as a "1" when the law or practice attempted to treat citizens in equal terms. In other words, when no distinctions were made based on religious grounds. If distinctions were made along religious lines this variable was then coded 87 "0". To illustrate, the case of Wheeler v. Barrera (1974) involved a federal program for educationally deprived stu- dents to be administered in both public and private schools, and so NEUTRAL was coded as a "1". On the other hand, it was coded as "0" in Lsmgp (1971) where only non- public schools were to be given funding. It was hypothesized in the last chapter that if a certain act or practice has a long history or tradition, the Court is far more likely to find it constitutionally acceptable. The reasons for this include the idea that such long term practices can become deeply embedded in the fabric of our national life and way of thinking. As part of this, it is possible that some practices which started out as being religious have over time ceased to have reli- gious meaning and have become secular in nature. Further- more, if a practice dates back to the time of the Framers of the Constitution, these practices must be acceptable since those who wrote the Establishment Clause did not cry out against them. From this hypothesis, the variable HISTORY-TRADITION was derived. This independent variable was coded 1/0. If the practice or activity had a long history dating back approximately two hundred years or more, HISTORY-TRADITION was coded "1". If this was not the case, it was coded "o" . be 88 The one questionable coding scenario which arose dealt with nativity scenes. In the two cases dealing with nativity scenes (Lynch v. Donnelly, 1984; pggpg_g§ Tppspees of the Village of Scarsgale v. McCrearv, 1985), no evidence was presented whether these particular types of displays go back to the time of the adoption of the Constitution. Instead, Chief Justice Burger stated in Lypgh only that such displays symbolized "a particular historic religious event, as part of a celebration acknowledged in the Western World for 20 centuries, and in this country by the people, by the Executive Branch, by the Congress, and the courts for 2 centuries..."(p.686). While no specific evidence was given dealing with the age of such displays, it seems quite possible that they do satisfy the two century requirement involved here. If not in terms of the exact date a nativity scene first appeared in a public square, then at least as part of a continued tradition of acknowledging Christmas as a national holiday or "celebration". For these reasons, HISTORY-TRADITION was coded "1" in these two cases. The fifth independent variable is LEVEL. This was derived form the hypothesis which stated that the Supreme Court is far more likely to accept practices involving colleges and universities or directed at adults. This related to the motivations of the particular institutions which were under scrutiny and how impressionable or sus- 89 ceptible to indoctrination the connected individuals might be. Ideally, this hypothesis would be operationalized in a way to incorporate every feature of it. However, decid- ing if a practice was directed at adults, at times, proved to be intractable. For example, is a nativity scene directed at adults or children? Therefore, the independ- ent variable LEVEL only concerns if the aid or activity involves a college or university. LEVEL was coded "1" if the decision involved institutions of higher education, and "0" otherwise. The sixth hypothesis stated that the Supreme Court will strike down any law or practice requiring governmental surveillance or significant inspection of religious institutions and their records. The justices claim they consider the resulting relationship between church and state when aid is proposed. If the relation- ship is too intimate or if the state must act as a watchdog there may be constitutional problems. SURVEILLANCE, the sixth independent variable, comes from this hypothesis. It is another 1/0 variable. SUR- VEILLANCE was coded as a "1" when there was "substantial" or "extensive" governmentally required reporting, regula- tions, on-site inspections, surveillance, or auditing of records; otherwise it was coded "0". In almost all cases where there was any type of the above mentioned require- ments attached to the aid or activity SURVEILLANCE was .F 9O coded "1" (e.g., Sandeps v. Johnson, 1971; New York y; Cgthsgral Academy, 1977; Eétats_of Thorton v. Caldor, 1985; School District of Grand Rapids v. Ball, 1985). Judgment was exercised in a few case where the con- tact or requirement was deemed to be so minimal as to not warrant such treatment (e.g., Walz v. Tax Commission, 1970; Tilton v. Richardsgp, 1971). To illustrate, in ngs, property used for religious purposes by religious organizations was granted tax exempt status. The law in question did not require any proof before such status was granted, it only needed to be claimed. Furthermore, even if the state later demanded proof (the law does not men— tion how this would be done) it seems quite possible that this would require very limited or cursory administrative contact in order to show that property was being used for religious purposes. Consequently, SURVEILLANCE was coded "0" in this case. The seventh hypothesis from the last chapter stated that if aid is given as a one-time grant rather than on a continuing basis to religious organizations, it is far more likely to be upheld by the Supreme Court. One major reason for this is the claim that annual or continuing appropriations to religious institutions are likely to lead to political divisiveness along religious lines. Such division is seen as a threat to the normal political process. From this hypothesis, the independent variable 91 ONE-TIME was drawn. It was coded in a very simple manner. If the aid was given only once, ONE-TIME was coded as a "1". In all other situations, it was coded "0". FREE EXERCISE is the eighth independent variable. It is derived from the hypothesis which stated that the Court would be somewhat more likely to decide a case in favor of a litigant stating a free exercise claim. As previously indicated, there is some literature which points to the Supreme Court favoring free exercise over establishment when the two conflict. Therefore, litigants who raise free exercise as an issue might be improving their chances of winning a case. FREE EXERCISE is coded in a slightly more complicated fashion than the previous seven independent variables which have been operationalized up to this point. Rather than being a 1/0 variable, FREE EXERCISE was coded as either "1" or "0" or "-1". It was coded as a "1" when free exercise was raised as an issue in the case and the litigant who raised the issue wanted an accommodationist decision by the Court. In other words, FREE EXERCISE was coded "1" when this issue was present in a case and when the party raising the issue wanted the outcome to be a "1" (the dependent variable coded as a "1"). As stated above, in most cases this meant that the party involved wanted a law or practice to be upheld and declared to be constitu- tional. If no free exercise claim was raised in a case, 92 then this variable was coded "0". If free exercise was stated as an issue, but the relevant party did not want an accommodationist decision (wanted the outcome to be a "0"), then FREE EXERCISE was coded as a "-1". The last of the nine hypotheses stated that if the U.S. is involved in an Establishment Clause decision, the Supreme Court is more likely to decide the case in favor of the position argued by the federal government. A num- ber of studies have shown that when the U.S. is a party to a case or an amicus, it has a relatively high success rate. The independent variable, US-USAMICUS, was drawn from this hypothesis. It has a similar coding scheme to that of FREE EXERCISE. If the U.S. was a party in a case or filed an amicus curiae brief and it argued for an accommodationist decision (wanted the outcome to be a "1"), US-USAMICUS was coded as a "1". If the U.S. was not a party in the case or an amicus then this variable was coded "0". Lastly, if the U.S. was a party or amicus in the case and it did not want an accommodationist decision (wanted the outcome to be a "0"), then US-USAMICUS was coded "-1". The final independent variables deal with the Court’s personnel. This research is based upon attitudinal and cognitive-cybernetic theory. It is thus being argued that it matters who is on the Court and how they perceive and react to the information given them. The justices have A. «0“. 1...; ‘. 93 goals and policy preferences and these affect the deci- sions they reach. In an attempt to take these factors into account, twelve "Justice" variables were added into the model. These variables are meant to tap into the specific biases and predispositions held by the individual justices. The twelve justice variables are: BLACK, DOUGLAS, STEWART, MARSHALL, BRENNAN, WHITE, BURGER, BLACKMUN, POWELL, REHNQUIST, STEVENS, and O’CONNOR. These variables represent twelve of the thirteen justices who served on the Court during Burger’s tenure as Chief Justice. One justice needed to be left off for computational reasons, and Justice Harlan was chosen.2 These variables were coded in a very simple fashion. A justice variable was coded "1" if the dependent variable, VOTE, was the vote of that particular justice. All eleven of the other justice variables were coded "0" for that decision or vote. For example, if the outcome to be predicted was Justice Black’s vote in a particular case, then the independent variable BLACK was coded as a "1" and the other eleven justice variables as "0". At most only one of these twelve independent variables can ever be coded "1" for any particular decision or vote. When adding these twelve justice variables with the nine factual variables previously elaborated upon, a com- prehensive model can be proposed to explain and predict 94 voting behavior in Establishment Clause cases. The model can be specified as follows: Yi= b0 + blxli + bzxzi + b3X35. + b4X4i + b5X51 b6X61 + b7X71 + bsxsi + b9X91 + bloxloi b11X111 + b12X121 + b13X13i + b14X14i b15X15i + b16X16i + b17X17i + b18X18i b19X19i + b20X201 + bZlXZIi + 91 -++-++ where Yi= VOTE (the vote of a justice in a particular decision) x1= PURPOSE X2= GENERAL GOVERNMENT SERVICE x3: NEUTRAL X4= HISTORY-TRADITION x5: LEVEL x6= SURVEILLANCE X7= ONE-TIME x8: FREE EXERCISE X9= US-USAMICUS X10: BLACK x11= DOUGLAS x12: STEWART x13= MARSHALL X14: BRENNAN X15: WHITE x15: BURGER x17= BLACKMUN X18: POWELL x19: REHNQUIST x20= STEVENS x21: O’CONNOR b0-21= coefficients ei= error term The anticipated influence of these variables should be clarified. In other words, what direction the coeffi- cients for the variables are expected to be in. GENERAL GOVERNMENT SERVICE, NEUTRAL, HISTORY-TRADITION, LEVEL, and ONE-TIME are all factual variables which indicate 95 instances where it has been hypothesized that the Court is more likely to act in an accommodationist fashion (which usually means to uphold a statute or program). Con- sequently, the coefficients for each of these is predicted to be positive. This is also true for the variables FREE EXERCISE and US-USAMICUS. In the case of these two vari- ables, when free exercise was raised as an issue or the U.S. government was involved in a decision, an accom- modationist outcome was predominately being sought by the relevant party. However, the opposite result is expected when either PURPOSE or SURVEILLANCE is present in a case. When these factual situations arise in a case, it was hypothesized that a separationist outcome would likely result. Thus, the coefficients for these two variables are expected to be negative.3 Based on the work of Harold Spaeth concerning the value systems of the justices (1979, pp.129-137) predic- tions about the justice independent variables can also be registered. As part of their value systems, Spaeth measured the support or nonsupport the justices showed for the value he dubbed "Freedom" (which included estab- lishment of religion cases). From this research the coef- ficients for variables BLACK, DOUGLAS, STEWART, MARSHALL, BRENNAN, and STEVENS are predicted to be negative. Whereas, the coefficients for WHITE, BURGER, BLACKMUN, 96 POWELL, REHNQUIST, and O’CONNOR4 are expected to be posi- tive. Lastly, it is quite important to identify the data that was used and the method of counting cases. The data set on which this research is based was primarily drawn from Dr. Harold Spaeth’s U.S. Supreme Court Data Base. Spaeth records every non-memorandum Supreme Court deci- sion, and includes "Issue Area" as one of his variables. Thus, from this data base one is able to identify Supreme Court decisions dealing with the Establishment Clause from the 1969-1985 terms. Once again, this time period was chosen in order to see if the justices’ decisions have been consistent under Chief Justice Burger and while sup— posedly relying on the purpose-effect-entanglement test. In addition to the 80 decisions drawn from Spaeth’s data base, 12 other cases were included in the data set. These were derived from two other sources. First, cases which were cited in the original 80 decisions were reviewed. It was concluded that 10 of these raised sig- nificant establishment of religion questions and should be included.5 Second, Lexis (a computerized legal text data base) was also used to search for Establishment Clause cases. It identified two cases which were not previously in the data set.6 As Rohde and Spaeth have written, "Although there is no inherent superiority in counting cases and votes one 97 way rather than another, the matter of method is suffi- ciently important to require specification"(1976, p.134). This statement is certainly true in this instance. The data set for this project involved considering 62 cases which dealt with the Establishment Clause and were given docket numbers by the Court during this time period. One unit of analysis for this project7 is each decision the Supreme Court reached in each of the 62 cases given a docket number. The reason for the distinction between decisions within cases and the cases themselves is because the Court, at times, makes several decisions within one case (e.g. Meek v. Pittenger, 1975; Wolman v. Walter, 1977). The Court may uphold certain parts of a statute and strike down other parts. Therefore, the Court often makes more decisions than the number of cases would indi- cate. Each decision was treated as a separate entity. This results in a data set of 92 "cases" or "observa- tions". Furthermore, these observations can also be broken down into the individual votes of the justices. There are, of course, nine potential votes in each decision. When individual voting is used as the unit of analysis, the data set consists of 790 observations (there are 38 "missing" observations due to less than nine justices par- ticipating in a decision). This individual level data set 98 of 790 observations will be analyzed extensively in the next chapter. B. Methodology The operationalized dependent variable in the model under study is not continuous. In fact, it is binary in nature and is associated with a qualitative choice made by the justices. The justices are faced with a choice between two alternatives (find there has been a violation of the Establishment Clause or there has not been one), and the choice they make depends on their own personal characteristics and preferences. It is assumed that there is an underlying scale, which is each justice’s propensity to vote one way or another in an Establishment Clause case. This underlying scale cannot be directly measured or observed, instead one merely knows how the justice actually voted in this binary situation. While there is a tendency for social scientists to automatically proceed with ordinary regression analysis in such a situation, this is not an appropriate strategy (Aldrich and Cnudde, 1975, p.579; McKelvey and Zavoina, 1975, p.103; Aldrich and Nelson, 1984, p.5). The reasons why regression analy- sis is inappropriate will be briefly stated, along with a discussion of the probit technique which will be used to estimate the parameters for this project. 99 The assumptions underlying the linear probability model require a continuous interval level dependent vari- able (Pindyck and Rubinfeld, 1981, p.273; Aldrich and Nel— son, p.9). Regression estimates with a qualitative binary dependent variable can result in serious errors in inference. John Aldrich and Forrest Nelson have stated that there are a number of unpleasant effects which result from an incorrect assumption of linearity (pp.27—30). For example, while estimates will tend to indicate the correct sign, these estimates have no known distributional proper- ties. These estimates may seriously misstate the mag- nitude of the true effects of the independent variables on the dependent variable. The probability estimates may even lie outside the range of 0 to 1. In other words, predictions of greater than 100% or less than 0% that an observation could have occurred. Simply put, the "regres- sion estimating procedure breaks down" (Aldrich and Cnudde, p. 579). With such serious problems connected with the linear probability model in such a situation, a nonlinear probability model will be used in its place. The nonlinear model is probit. Before using any eStimation technique, one should know what its underlying assumptions are. In the case of probit, there are four major assumptions (Aldrich and Nel- E .nnnu— wagon-1 ‘ 100 son, pp.48-49). The first assumption is that the depend- ent variable is binary. There are only two possible values it can take on. Second, since our concern is to interpret the dependent variable as the probability of making a choice, it is assumed that the use of the cumula- tive normal probability function will permit such an interpretation. Thus, unlike in regression analysis, we are not assuming that the dependent and independent vari- ables are linearly related. Third, it is assumed that the data are from a random samples, and that the observations on the dependent variable are statistically independent of one another (i.e., Y1, Y2, ...Yn are statistically inde- pendent). This is akin to assuming there is no serial correlation. The fourth assumption is that there is no exact or near linear dependencies among the independent variables. Each independent variable must have some variation across observations, and no two independent variables are perfectly correlated with one another. It should be stressed that as stated in assumption number two, probit is based upon the cumulative normal probability function (Pindyck and Rubinfeld, p.280). This ensures that the dependent variable will fall within the interval (0,1) for all values of the independent vari- ables. While ordinary least squares (OLS) regression results in predictions that may be interpreted as "probabilities", probit yields estimates that are true 101 probabilities and which lie in the appropriate range (Aldrich and Cnudde, p.581). Probit, consequently, is able to more accurately reflect the qualitative choice being made and how it has been measured. Probit is also a maximum-likelihood (MLE) technique (Aldrich and Nelson, pp.44-52). This contrasts quite sharply with OLS. While OLS is concerned with choosing estimates which yield the smallest sum of squared errors between predicted values and the data, MLE is concerned with picking estimates that imply the highest likelihood of having obtained the observed sample of the dependent variable. Aldrich and Nelson have written: "The princi- ple of MLE, quite simply, is to choose as an estimate of b that set of K numbers, say b’, which would make the likelihood of having observed this particular Y as large as possible"(p.50). As was already discussed, there are serious problems with regression estimates when there is an incorrect assumption of linearity. Therefore, the question becomes: what are the properties of MLE estimates? It turns out that these estimates exhibit asymptotic properties of unbiasedness, efficiency, and normality (pp.52-54). By being unbiased and efficient, this means the estimates are centered around the actual values on average, and no other unbiased estimator has a lower sampling variance. Since (7‘ " ; steal-€61- 102 these estimates possess the property of normality, this allows for hypothesis testing and inferences to be drawn. One also needs to consider what happens if the assumptions underlying this model are violated. Do the properties just discussed still hold? It is hard to ans- wer this question because there has been little work done in this area (p.81). It has been shown that when the residuals are serially correlated, MLE estimates remain unbiased in large samples, but they are not efficient and corrections for serial correlation have proven untractable (p.81). Thus, while there are no clear answers, if it appears that one or more assumptions have been violated, one should be quite wary of any inferences which are drawn. With this caveat in mind, it is necessary to discuss how to use and interpret probit estimates and measures. As stated above, probit estimates allow for hypothesis testing (Aldrich and Nelson, pp.54-55). This is intended to test whether an independent variable has a statisti- cally significant effect on the dependent variable. The t-statistic (the coefficient divided by the standard error) is used, just as it is with regression, to see whether the null hypothesis can be rejected. The t- statistic is compared with a one or two tailed critical value from the student’s t distribution with N-K (the num— ber of observations minus the number of independent vari- 103 ables) degrees of freedom and an a priori significance level. If the t-statistic exceeds the critical value, the coefficient is considered statistically significant, and the null hypothesis can be rejected. However, it should be noted that the interpretation of probit coefficients is not as straightforward as with regression coefficients (Aldrich and Cnudde, p.580). OLS coefficients can be interpreted as the change in the dependent variable which occurs with a one unit change in an independent variable. Whereas, probit coefficients represent the change in the z score for the cumulative normal probability function for a one unit change in the independent variable. Thus, if an OLS coefficient was .5, the change in the dependent variable which occurs with a one unit change in the relevant independent variable should also be .5. However, an equivalent .5 probit coef- ficient cannot be interpreted without first knowing the values of all of the other independent variables. Why this is the case will now be explained. With OLS the change in the dependent variable associ- ated with a one unit change in an independent variable should not be affected by the other independent variables. In fact, the interpretation of OLS coefficients is based on the assumption that all other values for the remaining independent variables are held constant (Pindyck and Rubinfeld, p.77). Furthermore, the change in the depend- -_P 104 ent variable is linear because the slope is constant. However, with probit the change in the dependent variable associated with a one unit change in an independent vari- able is affected by the values of the other independent variables. The degree of change is directly tied to them. This is because probit specifies a curvilinear relation- ship (which is S shaped). A constant change will have a variable effect on the probability of the dependent vari- able equalling 1 or 0. The effect of a constant change in an independent variable will have a greater impact on the dependent variable in the center of the cumulative normal probability function (where the probability is closest to .50) than in its tails. In other words, it matters where you start, and where you start depends upon the values of the other dependent variables. An example will be given to illustrate this point. Suppose the probit coefficient for the independent vari- able HISTORY-TRADITION is equal to 1.0. What effect does this have on the dependent variable? It, of course, also depends upon the other independent variables. Let us con- sider three scenarios where the values of all of the other independent variables indicate that there is a .10, .50, and .90 probability of the dependent variable (VOTE) equalling 1. This would be the result of their combined 2 scores equalling -1.28, 0.00, and 1.28. Where the probabiltiy had been .10 before HISTORY-TRADITION was pre- f' m'n-z" 105 sent, the probability jumps to .39 when it is added (a change of .29). In scenario number two, the probability jumps from .50 to .84 when HISTORY-TRADITION is present (a change of .34). In scenario number three, the probability jumps from .90 to .99 (a change of .09). This clearly indicates how the impact of an independent variable can vary depending upon the values of the other independent variables, and how necessary it is to have knowledge about all the variables before interpreting a probit coeffi- cient. In OLS regression analysis, an F statistic is used to test the significance of a model as a whole (tests the joint hypothesis that all the coefficients except the intercept are zero). A corresponding statistic based on the likelihood ratio principle is used in probit (Aldrich and Nelson, p.55). Also, the chi-square distribution is relied upon. The likelihood ratio statistic is compared to a critical value taken from a chi-square distribution table with K-l degrees of freedom and an a priori sig- nificance level to determine if the null hypothesis can be rejected. Lastly, the question of a goodness-Of-fit statistic needs to be addressed. Researchers are quite familiar with the coefficient of determination, R2, used in regres- sion analysis. With probit, McKelvey and Zavoina have claimed that while several statistics can be used to .e~ 106 measure the overall fit of a model, the most useful of these is the estimated R2 (1975, p.111). The estimated R2 is similar to the coefficient of determination in that it attempts to measure the portion of the original variance of the dependent variable explained by the independent variables. However, as the name implies, this is only estimated since there is no way of knowing the variance of the dependent variable on its underlying interval scale (p.111). Due to the estimation involved, both McKelvey and Zavoina, and Aldrich and Nelson state that an estimated or pseudo R2 should be used with caution (1975, p.112; 1984, p.59). Another much simpler measure of the overall fit of a model is the correct classification rate. This is the proportion of the time the model correctly predicts the observed value of the dependent variable. If the value of the dependent variable for an observation is "1", the model is considered to have correctly classified that decision if the independent variables have indicated that there is above a 50% chance of the dependent variable being a 1. Likewise, if the observed value is "0", the decision is treated as being correctly classified if the relevant probability if below 50%. Aldrich and Nelson do point out that when using this measure, one needs to be concerned with a baseline to compare the correct predic- tion rate against (p.57). Conclusions / Implications This proposed research combines previous work done in this field and adds to it. If the results of the estimation provide justification or corroboration for it, then it could be argued that this model has led to some progress in the field. This ,of course, depends on the ability of the model to explain and predict the Court’s decisions in this area of the law. One would need to con- sider the significance of the coefficients, the amount of variance explained (the R-Square), the percent of cases predicted correctly, the importance and relationship of the variables with one another, and potential problems with validity. There are, of course, a number of extensions to this work which could be looked into. For example, this model could be respecified and tested. While this research design only proposes to look at the Burger years, the Warren or perhaps the new Rehnquist Court could be studied. This model could also be used in an attempt to predict cases the Court has yet to hand down. These are just a few of the possibilities which could later be pursued. 107 Chapter Five: Presentation of the Results "On the other hand, some doubt has to be expressed about applying this method effectively to cases involving substantive civil rights, such as freedom of speech and freedom of religion . . . At the present stage of quantitative analysis, the concepts that appear in the context of substantive civil rights cases do not seem to be adaptable to quantitative interpretation."-- Fred Kort Up to this point, there has been a discussion of the problem to be addressed, a review of the literature in this area, and an explanation of the theory being espoused. From these elements hypotheses were derived and then operationalized. This chapter presents the results of" probit estimations ‘which. used. these operationalized variables. This will include a number of offshoots from the primary model being investigated. A. The Votes of all of the Individual Justices As discussed in the last chapter, the 92 decisions reached by the Burger Court concerning the Establishment Clause resulted in 790 votes cast by the individual justices. These votes were considered as either "accommodationist" (usually meaning that the law in question was considered to be constitutional and should be upheld) or "separationist" (usually meaning the law in question was considered to be unconstitutional and should be struck down). Accommodationist votes were coded as "1" 108 and separationist votes as "0". This resulted in 501 of the 790 votes being coded "O" (63%) and 289 being coded as "1" (37%). The justices most often voted in a separationist fashion holding that there had been a violation of the Establishment Clause. With this background, the results of the probit estimation of the Establishment Clause model, which was presented in the last chapter, are given in Table 1. As can be seen, it appears the model does a reasonably good job of explaining Burger Court decisions in this area. The estimated R2 turns out to be .67. The model also predicts 85% of the votes correctly. Since 63% of the votes result in a separationist decision, with no other information one could predict 63% of these decisions by simply choosing the modal category every time. However, by adding the independent variables given in the model there is approximately a 22% gain in the percentage of cases correctly categorized (or a reduction of error of approximately 60%). Also, the model as a whole, using the analog to the F-ratio, -2xLLR, is easily significant at the .005 level (with an -2xLLR value of 533.88). III—.— 110 Table 1 Probit estimation of the individual justices’votes Variable MLE S.E. MLE/S.E. PURPOSE - .37 .25 -1.47 GENERAL GOV. .44 .20 2.26* NEUTRAL 1.43 .15 9.35*** LEVEL .26 .24 1.07 HISTORY-TRADITION .95 .25 3.74*** SURVEILLANCE - .36 .15 -2.42** ONE-TIME 1.08 .29 3.77*** FREE EXERCISE .56 .13 4.14*** US-USAMICUS .52 .10 5.00*** BLACK - .34 .70 - .48 DOUGLAS -1.92 .71 -2.73*** STEWART - .01 .53 - .02 L MARSHALL -1.01 .54 -l.87* BRENNAN -1.09 .54 -2.03* WHITE 1.56 .52 3.02*** BURGER 1.24 .51 2.40** BLACKMUN - .40 .52 - .76 POWELL 1.65 .52 3.16*** REHNQUIST .08 .52 .16 STEVENS - .82 .56 -1.48 O’CONNOR .91 .57 1.59 CONSTANT -l.48 .51 -2.87*** Estimated R2 .67 . -2XLLR 533.88*** % correctly predicted 85.19 % in modal category 63.42 N 790 * significant at .05 ** significant at .01 *** significant at .005 In terms of the individual coefficients, all nine of the factual variables are in the predicted direction and eleven of the twelve justice variables are. Only the negative coefficient for BLACKMUN was not expected. This is not altogether suprising since Justice Blackmun was relatively a borderline case in Spaeth’s research and the 111 estimate found here is also relatively a weak borderline case. The significance of the coefficients was determined using a two-tailed t test for the variables FREE EXERCISE and US-USAMICUS, and a one-tailed t test for all of the remaining variables (since each could only take on values going in one direction). The t tests were based on the MLE/S.E. values given in the right hand column of Table 1. Of the factual variables, seven of the nine are significant at the .05 level. In fact, NEUTRAL, HISTORY- TRADITION, ONE-TIME, FREE EXERCISE, and US-USAMICUS are significant at .005. SURVEILLANCE is significant at .01. Yet, the coefficients for PURPOSE and LEVEL do not allow for the null hypothesis to be rejected at the .05 level of significance (PURPOSE is significant at .10). The reason these two variables are not significant at .05 may be the result of problems in operationalizing them. As was discussed in the last chapter, PURPOSE was the most difficult of all of the variables to operationalize, and was measured in what was deemed to be a very conservative manner. Furthermore, in operationalizing LEVEL part of its theoretical underpinnings were reluctantly unaccounted for in ordr to lessen the amount of subjective judgment. In terms of the typology given in Chapter Three, each category had at least one significant variable. Under "What is the aid", the variable GENERAL GOVERNMENT bangsgiiiaw I 112 SERVICES is significant at .05. Under "Who is receiving the aid", NEUTRAL and HISTORY-TRADITION are significant. Both of the variables falling under "What is the resulting relationship" are significant: SURVEILLANCE and ONE-TIME. Finally, both variables (FREE EXERCISE and US-USAMICUS) which are considered complicating issues are also significant. As for the justice variables, six of the twelve are statistically significant at .05. This includes DOUGLAS, WHITE, and POWELL which are significant at .005 and BURGER at .01. However, the coefficients for BLACK, STEWART, BLACKMUN, REHNQUIST, STEVENS, and O’CONNOR do not allow for the null hypothesis to be rejected at the .05 level (STEVENS and O’CONNOR are significant at .10). In terms of importance, variables with relatively large coefficients (in absolute value) have the potential to have the greatest impact on the outcome of a decision. They can potentially have the greatest influence on the estimated probability of whether the dependent variable takes on the value of 0 or 1. Thus among the factual variables; NEUTRAL, ONE-TIME, and HISTORY-TRADITION have the potential for the greatest impact (their respective coefficients are 1.43, 1.08, and .95). Among the justice variables, DOUGLAS, POWELL, WHITE, BURGER, BRENNAN, and MARSHALL potentially can have the greatest impact (their 113 respective coefficients are -1.92, 1.65, 1.56, 1.24,-1.09, and -1.01). However, as was explained in the last chapter, this influence depends upon the values of the other variables. This factor needs to be taken into account. For example, since the coefficient of HISTORY-TRADITION is .95, if the activity in question is historical or traditional, the probability of a justice voting to find the practice to be acceptable increases by .95 standard deviations (it adds .95 standard deviations to the cumulative probability function). If all other variables were controlled for (at z = 0.00, probability = .50), the probability of a justice finding an historical/traditional practice acceptable is .83. The probability of the activity not violating the Establishment Clause is thus 66% greater if the practice meets the requirements of this variable. Yet if all the other variables were controlled for at a higher level (z = 1.00, probability = .84), then the probability of a justice finding there is no violation goes up to .97. While the probability of finding no violation of the Constitution has increased, the impact of this variable is far less in this second scenario. It is quite likely that the practice would have been found to be constitutional even if it were not historical or traditional. With this idea in mind, Table 2 will now be presented. This table allows one to evaluate the changes 114 in probability associated with the addition of an independent variable when all other variables are controlled for in two ways. The first basic scenario given is when the probability of all other variables equals .50. This indicates that there is a 50% chance that the value of the dependent value will equal 1 before the addition of the relevant variable. The second scenario is controlling all other variables constant at their mean values before the addition of the relevant variable. The justice variables were not included in this second scenario. This is due to the fact that they are dummy variables whose average values are nonmeaningful. Only one can ever be present in any decision (their total averages add up to 1.0). Table 2 The impact of the estimates in two scenarios All other variables All other variables held at .50 probability held at their mean Variable MLE Starting Change in Ending Starting Change in Ending Prob. Prob. Prob. Prob. Prob. Prob. PURPOSE -.37 .50 -.14 .36 .25 -.10 .15 GEN. GOVT. .44 .50 .17 .67 .23 .15 .38 NEUTRAL 1.43 .50 .42 .92 .11 .47 .58 LEVEL .26 .50 .10 .60 .24 .09 .33 HISTORY .95 .50 .33 .83 .22 .35 .57 SURVETL -.36 .50 -.14 .36 .32 -.12 .20 ONE-TIME 1.08 .50 .36 .86 .23 .40 .63 FREE EXER .56 .50 .21 .71 .19 .19 .38 US-USAMICUS .52 .50 .20 .70 .21 .19 .38 BLACK -.34 .50 -.13 .37 DOUGLAS -1.92 .50 -.47 .03 STEWART -.01 .50 .00 .50 MARSHALL -1.01 .50 -.34 .16 L- 115 BRENNAN -1.09 .50 -.36 14 UNITE 1 56 50 .44 94 BURGER 1.24 .50 .39 .89 BLACKMUN - 40 50 -.16 34 POWELL 1 65 50 .45 .95 REHNQUIST .08 .50 .03 .53 STEVENS -.82 .50 -.29 .21 O’CONNOR .91 .50 .32 .82 When looking at the first half (variables held at .50) of Table 2, one notes that among the factual variables NEUTRAL, ONE-TIME, and HISTORY-TRADITION have the biggest impact. This, of course, is not very surprising based on their MLE values. It is interesting to see that their presence changes the odds of a justice finding a practice to be constitutional from 50% to above 80% in all three cases. The justices obviously take these factors quite seriously if they are present in a case. The importance of the often espoused benevolent neutrality doctrine appears to spring to the forefront with the ending probability for the variable NEUTRAL being .92. The justices seem to be far more willing to allow aid or a practice if the government is attempting to treat all of its citizens equally. It is also interesting to note that the complicating issues free exercise and the U.S. being involved in a case seem to have a significant effect. Both change the probability of finding a statute or practice acceptable by approximately 20%. Thus, it seems that the justices are influenced somewhat by these factors which lie beyond the primary question they are facing. 7'" t».—_——— _ 116 Likewise, it can be seen that the justice variables with the largest MLEs (DOUGLAS, POWELL, WHITE, BURGER, BRENNAN and MARSHALL) have the largest impact. But here three of these variables change the probability of the vote in a negative manner (DOUGLAS, BRENNAN and MARSHALL) and three in a positive manner (POWELL, WHITE and BURGER). This negative change means that their presence reduces the chance that the vote will be an accommodationist one. These justices are strongly predisposed to strike down most acts and practices. For example, if the variable DOUGLAS is added, the probability of an accommodationist outcome drops from 50% to only 3%. On the other extreme, one finds Justices Powell, White and Burger very predisposed to voting in an accommodationist fashion with their related variables having ending probabilities of .95, .94 and .89 respectively. The interpretation and importance of these coefficients should be taken a step further (and away from Table 2 for a moment). The dependent variable, VOTE, is the vote of a particular justice in a particular decision. For that observation, only one justice variable will be coded "1" (only one is present). It is the justice whose vote is the one the dependent variable is recording for that observation. Therefore, if VOTE is referring to Justice Douglas’ vote, then for there to be even a 50% chance of a predicted accommodationist outcome, the MLEs 117 for the factual variables must equal 3.4 (the MLE for DOUGLAS plus the constant equals - 3.4). It would take a combination of variables such as NEUTRAL, HISTORY- TRADITION and ONE-TIME to be present in a case (their MLEs sum to 3.46) for this to be the case. While Justice Douglas is the most extreme example, it does point out the importance of the justice variables since one is present in every case. This predisposition weighs most heavily when it is a large negative coefficient because of the additional presence of the negative constant. Furthermore, there is no guarantee of the factual variables with positive coefficients being present. In addition, only two of the justice variables have positive coefficients greater than the negative constant. Thus, these results indicate that there will be a predisposition in many observations to predict a "0" or separationist outcome before the factual variables are known, but this is clearly in accord with the fact that most votes are in that direction. This also shows up clearly in the second part of Table 2 (where the factual variables are held at their mean). As can be seen, all of the starting probabilities fall well below the .50 probability level. In fact, in only three cases (NEUTRAL, ONE-TIME and HISTORY-TRADITION) does the ending probability get over 50%. In six of the nine settings, even with the addition of the relevant .'Ir 118 factual variable one would expect a separationist or "0" outcome. Therefore, it becomes clear that in many situations it takes more than one variable with a positive coefficient to tip the balance in favor of an accommodationist decision. These results can also be broken down and considered in other ways. As was indicated earlier, the model predicted 85% of all cases correctly. However, as shown in Table 3, the model does better at predicting separationist or "0" outcomes than it does at predicting accommodationist or "1" outcomes. This is not totally unexpected when considering the negative predisposition just discussed. The model predicted or correctly classified 89.6% of the separationist decisions and 77.5% of the accommodationist ones. In either case this is a significant improvement over the modal result (63.42%). 119 Table 3 Crosstabulation of actual and predicted votes by category Predicted Count Row Row Pct 0 1 Total 449 52 501 0 89.6 10.4 63.4 Actual 65 224 289 1 22.5 77.5 36.6 Column 514 276 790 Total 65.1 34.9 100.0 These results can also be broken down justice by justice. How well did the model do at predicting the dependent variable when it referred to Justice Douglas’ vote as opposed to Chief Justice Burger’s for example? This breakdown is presented in Table 4. Table 4 Justice by justice breakdown N Modal for Correctly Justice (%) Predicted (%) HARLAN 14 64.29 100.00 BLACK 14 71.43 85.71 DOUGLAS 42 97.62 97.62 STEWART 64 71.88 87.50 MARSHALL 86 88.37 90.70 BRENNAN 90 90.00 90.00 WHITE 86 75.58 77.91 BURGER 87 65.52 72.41 BLACKMUN 87 80.46 91.95 POWELL 73 78.08 80.82 REHNQUIST 77 68.63 81.82 STEVENS 43 83.72 81.40 O’CONNOR 27 57.69 88.46 120 First, it should be noted that the percentage of correctly predicted observations for each justice beats the group modal percentage of 63.4%. In addition, these results pass even a tougher test. The percentage of correctly predicted observations in all but one case does at least as well and usually better when compared to the modal result for each individual justice. The justices differ in how they vote and consequently their individual modal categories are quite different (it should also be mentioned that modal categories can indicate a majority of either "0" or "1" votes for different justices). The model does slightly worse at predicting Justice Stevens’ votes than the modal outcome would (81.40 compared to 83.72%), but in all other cases it at least ties its predictive accuracy. In fact, it beats it for 10 out of the 13 justices. This is more impressive when one considers how high the modal category is for a number of the justices. Thus, the primarily model of this research is successful in a number of ways. First, it does a rather good job of explaining the votes of the individual members of the Burger Court when taken as a group (correctly predicting 85%). Second it is able to predict votes which fall into either the "0" or "1" category at well above the modal rate. Third, the model in most cases performs I. . 121 better than the modal category when the results are broken down justice by justice. With these supportive overall results, other interesting aspects of this work can now be looked into. B. Reduced Model What happens to the results if only factual variables are included in the model? Or, what would happen if only the justice variables were incorporated? In this next section, these two specifications of the model will be considered. Table 5 displays the results of a fact variable only version of the model. Once again, seven of the nine variables are statistically significant at .05 (this includes a number of the estimated coefficients significant at .005). The model as a whole is also significant (at .005). However, this reduced version of the primary model has an estimated or pseudo R2 of only .27. More importantly, the percentage of correctly predicted cases drops to 72.66%. 7‘ 122 Table 5 Fact variable version of model VARIABLE MLE S.E. MLE/S.E. CONSTANT -0.98 0.15 -6.17*** PURPOSE -0.12 0.21 -0.58 GEN.GOVT. 0.35 0.16 2.19* NEUTRAL 0.97 0.11 8.50*** LEVEL 0.22 0.21 1.04 HISTORY 0.69 0.22 3.19*** SURVEIL. -0.21 0.12 -1.73* ONE-TIME 0.69 0.23 2.99*** FREE EXER. 0.44 0.11 4.10*** US-USAMICUS 0.36 0.09 4.16*** Estimated R2 .27 -2XLLR 218.16*** % correctly predicted 72.66 % in modal category 63.42 N . 790 * Significant at .05 ** Significant at .01 *** Significant at .005 These goodness of fit statistics indicate that this revised version of the model has lost a significant portion of its explanatory power. This specification is still able to beat the modal category of 63.42% and correctly predict a fairly large portion of the votes. It lends credence to the theory being espoused in this research concerning the importance of simple cues or signals to the justices. However, it is clearly missing other important factors of the decision making process. Table 6 gives the results for the model when it only includes the justice variables. As with the full model f"____‘T“' 123 six of the twelve variables are significant at .05 (with DOUGLAS, WHITE and POWELL significant at .005). The model as a whole is significant at .005. Once again, there is a drop in the estimated R2 and the percentage of correctly predicted cases in this reduced version of the primary model. Yet, this drop is not as large as with the fact variable only model. Here the estimated R2 is .39 and the percentage of correctly classified cases is 78.10%. Table 6 Justice variable version of model VARIABLE MLE S.E. MLE/S.E. CONSTANT -0.43 0.33 -1.29 BLACK -0.14 0.49 -0.28 DOUGLAS -1.55 0.54 -2.89*** STEWART -0.15 0.37 -0.40 MARSHALL -0.76 0.38 -2.01* BRENNAN -0.85 0.38 -2.24* WHITE 1.12 0.37 3.07*** BURGER 0.83 0.36 2.29* BLACKMUN -0.43 0.37 -1.16 POWELL 1.21 0.37 3.24*** REHNQUIST -0.06 0.37 -0.16 STEVENS -0.55 0.41 -1.36 O’CONNOR 0.63 0.42 1.50 Estimated R2 .39 -2XLLR 313.88*** % correctly predicted 78.10 % in modal category 63.42 N 790 * Significant at .05 ** Significant at .01 *** Significant at .005 These results again indicate that there is a loss in explanatory and predictive ability. However, the loss in 124 this second specification is not as great. This justice variable only model also provides support for the theory being espoused here. Based on judicial behavior research, the importance of the predispositions of the justices can hardly be overstated. It matters who is on the Court and what attitudes and values these individuals hold. The importance of these predispositions in predicting outcomes was stated in the last section. After comparing these two models it appears that the justice variables may carry somewhat more weight than the fact variables in correctly predicting voting outcomes. However, as should be clear, both models indicate that the variables which were incorporated into them are useful in explaining and predicting these decisions. Either model is a useful tool for looking at these cases. Yet, neither model is as good as the full model . This is, of course, completely in line with what has been hypothesized in this research. While the biases and values of the justices are extremely important, there are cognitive limitations which can influence decisions. Therefore, both of these factors should be looked into when studying Supreme Court decision making. C. Ths Justices One by One In Section A of this chapter, the findings for the primary model were broken down to consider how well it did 125 justice by justice. This was done by separating the 790 observations by the justice who voted in that particular case and seeing what percentage were correctly predicted. These findings indicated that the primary model beat or tied the individual modal category for twelve of the thirteen justices. In this section, the individual justices will be examined more closely. Here the fact variable version of the model was run on each of the individual justices. The justice dummy variables were excluded since the model is only attempting to predict the vote of one justice. This analysis is an attempt to not only see how well the model predicts the votes of the individual justices when they are separated, but also which factual variables are important to each. For example, do Justice Marshall and Powell focus in on different facts in a case? Unfortunately, a number of methodological problems intervene in attempting to do this. The primary problem is simply too few cases having been decided. When the 828 potential observations of the full data set are broken up by justice, the largest possible number of votes by any justice is 92 (828 divided by 9). No justice even reaches that figure. Brennan leads the way with 90 decisions. At the other end of the range, Harlan and Black were only involved in 14 decisions. :T“ 126 Complicating this problem even further, is the fact that with a number of the justices there is little variance to explain. For example, the modal categories for Justices Blackmun, Brennan, Marshall, Stevens and Douglas are all over 80%. Douglas ,in fact, voted 98% of the time in a separationist direction. Additionally, extreme voting patterns can lead to facts that are perfectly correlated with a justice’s decisions. In other words, every time a certain fact situation was present a justice voted in a certain direction in a case. When this happens a probit model cannot be estimated. With so few cases, a single exception can often dramatically affect an estimate, or even allow the model to be estimated. Due to these problems, the fact variable version of the model could not be estimated for Justices Harlan, Black, Douglas and O’Connor. These are the four justices involved in the least number of decisions (ranging from 14 to 42). Furthermore, the estimated R2 will not be presented for the remaining nine justices since it appeared to be very unreliable. It fluctuated wildly from .17 to the high .90’s. With these exclusions and caveats in mind, the results are provided in Tables 7 - 15. We 127 Table 7 Justice Stewart VARIABLE MLE S.E. MLE/S.E. CONSTANT -1.17 0.79 -1.48 PURPOSE 4.43 22.78 0.19 GEN. GOVT. 1.53 0.68 2.24* NEUTRAL 1.24 0.57 2.18* LEVEL 1.47 1.22 1.21 HISTORY 4.06 14.74 0.28 SURVEIL. -1.10 0.57 -1.91* ONE-TIME 0.39 1.20 0.32 FREE EXER. 0.32 0.61 0.53 US-USAMICUS 0.17 0.38 0.44 -2XLLR 54.08*** % correctly predicted 89.06 % in modal category 71.88 N 64 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 Table 8 Justice Marshall VARIABLE MLE S.E. MLE/S.E. CONSTANT -6.00 13.89 -0.43 PURPOSE -1.22 28.92 -0.04 GEN. GOVT. -0.12 0.96 -0.13 NEUTRAL 5.29 13.89 0.38 LEVEL -0.30 1.18 -0.26 HISTORY 3.44 13.88 0.25 SURVEIL. -1.09 0.91 -1.19 ONE-TIME -1.88 42.00 -0.04 FREE EXER. 2.19 0.91 2.41* US-USAMICUS -0.07 0.44 -0.16 -2XLLR 98.23*** % correctly predicted 95.35 % in modal category 88.37 N 86 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 128 Table 9 Justice Brennan VARIABLE MLE S.E. MLE/S.E. CONSTANT -6.52 25.06 -0.26 PURPOSE -1.67 60.76 -0.03 GEN. GOVT. -4.21 62.98 -0.07 NEUTRAL 6.06 25.07 0.24 LEVEL -0.60 1.23 -0.48 HISTORY 4.01 25.06 0.16 SURVEIL. -1.37 1.08 -1.27 ONE-TIME -1.90 81.16 -0.02 FREE EXER. 2.16 0.94 2.30* US-USAMICUS -0.42 0.58 -0.72 -2XLLR 109.36*** % correctly predicted 95.56 % in modal category 90.00 N 90 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 Table 10 Justice White VARIABLE MLE S.E. MLE/S.E. CONSTANT -1.10 0.59 -1.87* PURPOSE -0.72 0.70 -1.03 GEN. GOVT. 1.32 0.77 1.72* NEUTRAL 1.74 0.61 2.85*** LEVEL -l.01 0.82 -1.24 HISTORY 1.76 1.01 1.74* SURVEIL. 1.04 0.53 1.96* ONE-TIME 1.64 0.96 1.71* FREE EXER. 0.67 0.37 1.80* US-USAMICUS 1.55 0.54 2.89*** -2xLLR 60.90*** % correctly predicted 81.40 % in modal category 75.58 N 86 + significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 129 Table 11 Justice Burger VARIABLE MLE S.E. MLE/S.E. CONSTANT -0.61 0.49 -1.25 PURPOSE -0.16 0.63 -0.25 GEN. GOVT. 0.14 0.54 0.26 NEUTRAL 1.67 0.42 3.93*** LEVEL 0.17 0.88 0.20 HISTORY 0.70 0.75 0.93 SURVEIL. 0.06 0.46 0.13 ONE-TIME 4.40 15.61 0.28 FREE EXER. 0.46 0.30 1.50+ US-USAMICUS 0.84 0.33 2.54** -2XLLR 40.66*** % correctly predicted 77.01 % in modal category 65.52 N 87 + significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 Table 12 Justice Blackmun VARIABLE MLE S.E. MLE/S.E. CONSTANT -2.66 1.16 -2.28* PURPOSE 1.05 1.05 1.00 GEN. GOVT. 1.01 0.79 1.29 NEUTRAL 2.13 0.97 2.20* LEVEL 2.34 1.02 2.30* HISTORY 1.35 1.01 1.33 SURVEIL. -1.06 0.62 -1.71* ONE-TIME 3.29 1.31 2.51** FREE EXER. 0.42 0.71 0.59 US-USAMICUS 0.66 0.38 1.74* -2xLLR 43.40*** % correctly predicted 91.95 % in modal category 80.46 N 87 + significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 130 Table 13 Justice Powell VARIABLE MLE S.E. MLE/S.E. CONSTANT 0.67 0.67 0.99 PURPOSE -1.08 0.83 -1.30+ GEN. GOVT. -1.02 0.65 1.59+ NEUTRAL 1.04 0.56 1.86* LEVEL 3.08 33.67 0.09 HISTORY 4.96 34.16 0.15 SURVEIL. 0.24 0.52 0.47 ONE-TIME 4.36 38.40 0.11 FREE EXER. -0.73 0.55 -1.32+ US-USAMICUS 1.00 0.45 2.26* -2XLLR 52.10*** % correctly predicted 87.67 % in modal category 78.08 N 73 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 Table 14 Justice Rehnquist Variable MLE S.E. MLE/S.E. CONSTANT -O.84 0.65 -1.29 PURPOSE -0.84 0.84 -1.00 GEN. GOVT. 1.27 0.59 2.16* NEUTRAL 1.16 0.46 2.55** LEVEL 1.19 0.74 1.61+ HISTORY 1.15 0.74 l.55+ SURVEIL. -0.99 0.48 -2.07* ONE-TIME 1.03 0.88 1.17 FREE EXER. -0.20 0.58 -0.34 US-USAMICUS 0.42 0.32 1.35+ -2xLLR 48.47*** % correctly predicted 84.42 % in modal category 68.83 N 77 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 131 Table 15 Justice Stevens VARIABLE MLE S.E. MLE/S.E. CONSTANT -4.39 27.92 -0.16 PURPOSE -4.09 50.37 -0.08 GEN. GOVT. 0.01 1.06 0.01 NEUTRAL 0.27 36.03 0.007 LEVEL 0.06 29.10 0.002 HISTORY -3.75 24.51 -0.15 SURVEIL. -7.29 26.82 -0.27 ONE-TIME 3.95 136.89 0.03 FREE EXER. 8.07 28.07 0.29 US-USAMICUS 3.68 16.26 0.23 -2XLLR 49.20*** % correctly predicted 93.02 % in modal category 83.72 N 43 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 The overall results for the individual models are clearly more problematic than those when the justices are pooled together. Yet, there are a number of bright spots. First of all, each of the nine models as a whole are easily significant at .05. Secondly, each model does a good job of correctly predicting the voting decisions of the justices. Each is able to beat the relevant modal category for that model. Once again, this is relatively difficult since the modal categories for some of the justices are quite high. The models were able to correctly predict between 77% (for Justice Burger) and 96% 132 (for Justice Brennan) of the votes cast. The predictive accuracy improved for Justice Stewart from a base of 72% up to 89% (a 61% reduction in error), for Justice Blackmun from a base of 80% up to 92% (a 60% reduction in error), and for Justice Rehnquist from a base of 69% to 84% (a 48% reduction in error). Yet, while the predictive accuracy for the models was quite high most of parameter estimates did not turn out to be statistically significant at the .05 level. The results for Justice White were the best on these grounds with seven of the nine variables being significant (the same number as in the pooled justice data set). At the other extreme, were the results involving Justice Stevens in which no variable was significant. Which factors were most important to which justices? For Justice Stewart, GENERAL GOVERNMENT SERVICES, NEUTRAL and SURVEILLANCE appear to be the most important. In his case an activity which would have a 50% probability of being allowed has over a 98% probability if it can be considered a general government service or if it treats all groups in a neutral fashion. However, if surveillance is involved a .50 probability drops to approximately .03. In Justice Marshall’s case it is significant whether free exercise is an issue. He is far more likely to vote in an accommodationist fashion if this complicating issue is present. Marshall’s record indicates that he voted 88% 133 of the time in a separationist manner. But, if all other variables were controlled for at only a 10% probability of allowing an activity and free exercise was also present, the probability of voting in an accommodationist direction would rise to approximately 87%. Table 9 shows that this same scenario is also true for Justice Brennan. His voting record stands at 90% separationist votes. However, when free exercise is present as an issue 2.3 standard deviations are added to the cumulative probability function which can have a dramatic effect on a voting decision. Justice White, as mentioned above seems to be affected by the same variables that were deemed significant when all of the justices’ votes were pooled together. There are differences, of course, between White’s results and the pooled results. Two interesting differences are that the largest estimated coefficient for White belongs to HISTORY-TRADITION not NEUTRAL, and White also places more emphasis upon whether the U.S. is involved in a case. For Justice Burger, NEUTRAL and US-USAMICUS are clearly the most influential variables. While no other variable is significant at .05, both of these variables are significant at .01. Like Justice White, Burger seems to be affected by the presence of the U.S. in a case. This is not surprising since Burger has often advocated 134 deference to other branches of the government. Of course, there are other possible explanations for why Burger seems to be influenced by this factor. Table 12 presents the results for Justice Blackmun. Five variables turn out to be significant in his case. They are NEUTRAL, SURVEILLANCE, ONE-TIME and US-USAMICUS. It should be noted that this is the only instance in this research where LEVEL was found to be significant at the .05 level. In fact, it has a larger coefficient than NEUTRAL. Blackmun appears to closely consider the differences between colleges and universities as opposed to high schools and elementary schools. Justice Powell, along with White, Burger, and Blackmun is significantly influenced by the presence of the U.S. in a case. He also places importance upon whether the situation can be deemed to be neutral in its treatment of different groups. It will also be pointed out that three other variables (PURPOSE, GENERAL GOVERNMENT SERVICES and FREE EXERCISE) while not significant at .05 are significant at .10 here. Table 14 displays the results for Justice Rehnquist. He along with Justices Stewart, White, Burger, Blackmun and Powell appear to be significantly influenced by NEUTRAL. This variable is thus the most frequently significant variable among the group. Additionally, importance is placed upon GENERAL GOVERNMENT SERVICES and 135 SURVEILLANCE by Rehnquist. It will also be pointed out here that there are three other variables (LEVEL, HISTORY- TRADITION and US-USAMICUS) which are not significant at .05 but are at .10. Lastly, there is Table 15 which presents the probit estimates for Justice Stevens. As discussed above, none of these estimates are significant at .05. Justice Stevens’ by far was involved in the smallest number of cases (43). In ending this section, it will once again be mentioned that due to the small number of cases the estimates are somewhat less reliable than one would hope for. As has been stated, only a few cases could dramatically change these estimates. This is not only true in the Stevens’ case but for some of the justices as well. Thus, these results should be treated with a certain degree of caution. D. Decisions by the Court In the final section of this chapter, the fact variable version of the Establishment Clause model will again be estimated. But unlike all of the previous estimations, this one will involve the decision made by the entire Supreme Court, (not simply the voting decision of one of the justices). In other words, the dependent 136 variable here is the outcome or decision reached by the entire Court. Thus, it requires a majority of the justices to determine if it is a "0" or "1" outcome. Up to this point, it has been shown that the full model does a good job of explaining and predicting the pooled votes of the justices. When the fact variable version of the model is run for each justice, overall the results are less impressive. However, the predictive accuracy for the nine individual justice models was quite high. The final question to be asked is how well the model does when the votes of the justices are aggregated. For society this is the most important result. It does not matter how a particular justice voted but instead what decision was reached by the Court as a whole. In order to test the model’s~ability to explain and predict the Court’s decisions the individual votes of the justices were simply added up. If the majority of the justices voted "0" in a case then the dependent variable was coded "0". Likewise, if the majority voted "1" then the dependent variable was coded "1". This, of course, leads again to a reduction in the number of observations. Since there are 92 decisions the Court considered, this is number of observations in this data set. The results of this estimation are displayed in Table 16. As can be seen the model as a whole is easily significant at .05. Additionally, the percent of F 137 correctly classified decisions is quite high at 88%. It should be pointed out, however, that the modal category for Court decisions is 68%. Table 16 Decisions by the Court VARIABLE MLE S.E. MLE/S.E. CONSTANT -0.96 0.60 -1.62+ PURPOSE -0.72 0.81 -0.89 GEN. GOVT. 1.44 0.58 2.48** NEUTRAL 1.36 0.42 3.22*** LEVEL 1.13 0.73 1.53+ HISTORY 1.11 0.68 1.65+ SURVEIL. -1.18 0.46 -2.55** ONE-TIME 1.17 0.86 1.35+ FREE EXER. 0.14 0.47 0.30 US-USAMICUS 0.34 0.30 1.12 -2XLLR 69.91*** % correctly predicted 88.04 % in modal category 68.48 N 92 + Significant at .10 * Significant at .05 ** Significant at .01 *** Significant at .005 Once again with a lower number of cases there are more problems with the parameter estimates. In this situation, three of the nine fact variable are significant at .05. (GENERAL GOVERNMENT SERVICES, NEUTRAL and SURVEILLANCE). Surprisingly, GENERAL GOVERNMENT SERVICES (MLE = 1.44) has the potential to be slightly more influential in the Court’s decision than does NEUTRAL (MLE = 1.36). After this first group of highly significant AF— AMA'I-M" .10 (HISTORY-TRADITION, LEVEL and ONE-TIME). Another interesting result here is that while both FREE EXERCISE and US-USAMICUS are significant in the pooled individual justice estimation, neither is close to .05 significance here. Table 17 provides a breakdown of the results by separationist or "0" outcomes and accomodationist or "1" outcomes. AS can be seen the model does better at predicting separationist outcomes than it: does accommoditionist ones. The model predicted or correctly classified 95% of the separationist decisions and 72% of the accommodationist ones. In both cases there is an improvement over the modal result (68%). While these results fall in line with those discussed earlier for the full model, there is a greater difference here between the two categories. The higher modal percentage of separationist decisions and a small number of accomodationist decisions (29) could account for this difference. Table 7 Crosstabulation of Actual and Predicted Votes by Category Predicted Count Row Row Pct O 1 Total 0 60 3 63 95.2 4.8 68.5 Actual 1 8 21 29 27.6 72.4 31.5 Column 68 24 92 Total 73.9 26.1 100.0 137a Thus, while these results are not as jpleasing as those for the full model, overall they can bee seen as encouraging and helpful. These findings (particularly the high predictive accuracy rate) provide further support for some of the ideas being tested here. From a different angle, they once again show the importance that particular cues or facts can have on decision making on the Supreme Court. 138 Chapter 6: Summary and Conclusion "It can be seen, therefore, that the greatest need for mathematical analysis exists not in the ’settled’ areas of the law, but in those areas where the law is ’unsettled’..." - Fred Kort. This work began with a discussion of U.S. Supreme Court decisions dealing with the Establishment Clause of the Constitution. It was shown that over the years the Court had claimed to rely on a number of tests and doctrines. Then in 1970 in the first Establishment Clause case decided during Warren Burger’s tenure as Chief Justice, the Court laid down the third part of the present three-part Establishment Clause test (the purpose-effect— entanglement test). Yet, the Court’s use of this test did not clear up this issue area for many scholars. For example, it has been stated that the Supreme Court has not provided clear guidance, but rather has provided more questions than answers. Constitutional law is this area is confused, conflicting and uncertain. These decisions are ad hoc judgments which are incapable of being reconciled on any principled basis. Furthermore, Establishment Clause cases have become totally unpredictable. Thus, this area of law appeared to be the type of "unsettled" law that Fred Kort refers to in the above quotation. An area of law which most scholars see as 139 140 being inconsistent, incoherent, or unclear. If order or consistency can be shown to occur with such a group of cases, this is a much greater achievement than in a more "settled" area of the law. Since the primary goal of this research is to show that the behavior of Supreme Court justices can be explained, predicted and understood, this area of law was selected to provide a rigorous test. In addition to this larger goal of explaining Supreme Court behavior in general, this research could also simultaneously attempt to show that the Establishment Clause decisions are far more consistent than scholarly opinion would have one believe. With these goals and background, the literature relevant to this study was reviewed. It includes traditional legal scholarship, judicial behavior and attitudinal research, and fact or cue models. This dissertation, to some extent, attempts to tie these different scholarly approaches together. Primarily it suggests the integration of attitudinal theories with cue or fact theory. This research also provides a more elaborate theoretical basis for cue theory. Spelled out briefly, the theoretical perspective relied upon here is based on David Rohde and Harold Spaeth’s framework for explaining Supreme Court decisions. They have written that Supreme Court outcomes are the consequence of three factors: goals, rules and 141 situations. First of all, Rohde and Spaeth assume that justices have certain goals they wish to achieve. These goals are policy goals. Thus, one needs to take into account the policy preferences of the justices (which are based upon the individual’s beliefs, attitudes and values). The second factor is the rules of game. However, Rohde and Spaeth point out that for the justices these rules are not usually very constraining. The third factor of the framework is the situation facing the Court. This third factor is the linchpin for tying together Rohde and Spaeth’s work with cue theory and bounded rationality or cognitive-cybernetic theory. Following the work of Herbert Simon and John Steinbruner (among others), this research also assumes that due to the complexity and uncertainty of the environment, and the limitations of the human mind, decision makers are forced to act in a boundedly rational or cognitive-cybernetic fashion. Simon defines bounded rationality as rationality in situations where the complexity of the situation is immensely greater than the computational powers of the adaptive system. Closely tied to this, Simon claims that individuals will often behave in a "satisficing" manner meaning they will be content with a "good enough" solution or course of action. Similarly, cybernetic theory argues that decision makers will attempt to hold uncertainty to a minimum. Instead of .0“. 142 considering all of the possibilities and facts in a situation, individuals attempt to control variety and uncertainty by means of highly focused attention and an established set of responses. In other words, uncertainty and variety are reduced because only a few critical variables and possible responses receive attention. The conclusion reached in this research is that due to the amount of work before the Supreme Court and its complexity, the justices must often rely upon a simple decision making structure. In almost every case before them, there are countless facts or factors which could be considered. However, the justices arguably do not have the time, resources or intellectual capacity to consider every feature of every case. Instead, it is argued that the justices are often forced to behave in a boundedly rational or cognitive-cybernetic fashion. Only a few critical cues or facts are focused on. The justices also rely upon previous experience and decision rules to simplify the decision they need to make. Therefore, when hearing an Establishment Clause case, certain facts will stand out as being important. Upon accepting the idea that there are key facts or cues which guide the justices, the next question becomes: What are the important facts or cues in Establishment Clause cases? This study proposes the use of a typology or classification scheme in order to answer this question. 143 The typology comes from the Lsmon v. Kurtngp decision, and it consists of three basic questions. First, what is the nature or purpose of the aid/practice being proposed? Second, who is getting the aid or will be affected by this practice? Third, what is the resulting relationship between government and religion? It is claimed in this research that these three basic questions cover the key and fundamental issues facing the Court in this area. Additionally, it was argued that two other complicating issues which could influence the outcome of a case should be considered. With this typology as a guide, specific aspects of each major heading were sought after. From this search, nine testable propositions were put forth. First, it was hypothesized that where the sole (or predominant) purpose of a law is religious, the Court will find the law to be unconstitutional. Second, where the aid fits the description of being a nonideological, general, welfare service, the Court will find it to be constitutional. Third, it was hypothesized that when legislation aids or affects all groups equally, it has a far greater chance of being upheld. Fourth, if a certain act or practice has a long history or tradition, the Court is far more likely to find it constitutional. Fifth, the Supreme Court is far more likely to accept practices involving colleges and universities or directed at adults. Sixth, it was 144 hypothesized that the Supreme Court will strike down practices requiring governmental surveillance or significant inspection of religious institutions and their financial records. Seventh, if aid is given as a one-time grant it is far more likely to be upheld by the Supreme Court. Eighth, if free exercise is also an issue in a case, the Court will be somewhat more likely to decide a case in favor of the party making the free exercise claim. Ninth, if the U.S. is involved in an Establishment Clause decision, it is hypothesized that the Court is more likely to decide the case in favor of the position argued by the U.S. These nine hypotheses were then operationalized and an Establishment Clause model constructed. The data used to test the model were the Establishment Clause decisions of the Burger Court (1969 through 1985 terms). Probit was used to estimate the parameters of the primary model (and versions of it). The results of the empirical analysis provided general support for what has been argued here. The primary model’s performance was rather good. The model as a whole was highly significant, it had a reasonably high pseudo R2, and it correctly categorized 85% of the justices’ votes. It was able to predict votes in either the "0" or "1" category at well above the modal rate. The model in most cases also performed better than the modal 145 category when the results were broken down justice by justice. In terms of the estimated coefficients, most were significant at .05, and 20 of the 21 were in the predicted direction. NEUTRAL, ONE-TIME, and HISTORY- TRADITION turned out to be the fact variables which had the greatest potential influence. Meanwhile, PURPOSE and LEVEL were the most disappointing since they did not reach the acceptable level of significance. The results were quite compatible with previous attitudinal research. The estimated coefficients for the justice variables clearly indicated the importance of the predispositions of the individual justices. Likewise, when the primary model was reduced into a fact variable only model or a justice variable only model, the latter seemed to be slightly more effective at explaining the results of these decisions. But it should be made clear that while both reduced models did an adequate job in dealing with these cases, neither was as good as the full model. Less pleasing were the results dealing with the nine individual justice models. While each model was statistically significant as a whole, most of the parameter estimates were not at the .05 level. This examination was clearly hindered by the low number of cases in each justice’s data set and the extremeness of some of the justices’ voting patterns. Even with these 146 problems, the models were still able to correctly categorize between 77% and 96% of the votes cast. The final part of the analysis involved the decisions of the Court as a whole. This resulted in a reduced data set of only 92 cases. Once again, the model as a whole was significant at .05, but only three of the nine variable were (six of the nine were at .10). This model was also able to achieve a rather high correct classification rate of 88%. In conclusion, this research appears to do a rather good job of explaining the Court’s decisions in this area. The individual justices seem to be predisposed in certain directions, and there are certain facts which seem to greatly influence them. These cases are not as incoherent and confused as some scholars have stated. In fact, when viewed from the perspective used here, most of the decisions can be explained and predicted in a consistent manner. Thus, an area that many have thought was unsettled was shown to have a good deal of order. In addition to attempting to explain this area of law, this research hopefully also contributes to explaining and predicting Supreme Court decision making in general. This research builds upon previous judicial behavior theory and supplements it with work dealing with decision making limitations. The overall results of the analysis provide justification or corroboration for this 147 theoretical perspective, and thus it can be argued that this work has led to some progress in the field. There are, of course, a number of extensions to this work which could be looked into. For example, while this project looked at the Burger Court years, the new Rehnquist Court could be studied. The model used here (with some adjustments for the new justices on the Supreme Court) could be used in an attempt to predict cases the Court has yet to hand down. Possibly other additional influential factors could be identified and added to the model, or improvements made in how the present variables are operationalized (e.g. PURPOSE and LEVEL). Lastly, the ideas presented here should be tested in a number of other areas of law and on courts other than the U.S. Supreme Court. Clearly there is much more work to be done. Notes Introduction 1The 1968 Board of Education v. Allen case should also be mentioned. In this case the Supreme Court upheld a statute providing for the loan of secular textbooks authorized for use in public schools to pupils attending nonpublic schools. The Court relied upon both Everson and the secular purpose and effect tests in making its decision. The majority opinion stressed that the financial benefit of this program was to the parents and children rather than to the schools. This parental or child benefit criterion has not been used by the Court since Allen in any regular or consistent manner. Chapter 3 11t should be noted, however, that there are occasions when the Court’s power and authority are threatened. When this occurs, the justices may feel pressure to vote in a direction other than the one their personal preferences would dictate (for more detail see Murphy, 1964; Ulmer, 1971; Rohde, 1972; Hutchinson, 1979). 2These three "evils" are also cited by the Court in Walz v. Tax Commission, 1970; Tilton v. Richardson, 1971; Com- mittee for Public Education v. NVguist, 1973; Meek v. Pit- tenger, 1975; Lynch v. Donnelly, 1984; and School District of Grand Rapids v. Ball, 1985). 3The first two parts of the test were adopted from deci— sions prior to Burger’s appointment to the Court. The third requirement was added in Burger’s very first Estab- 1ishment Clause case, Walz, in which he wrote the majority opinion. 4All of the fOllowing quotations are taken from majority Opinions or from opinions concurring with the majority. Chapter 4 1The word "primary" is used because while the focus of this research project is on the voting patterns of the individual justices, the decisions reached by the entire Court will also be considered in Chapter Five. 2As 'will be seen Shortly, ‘these justice 'variables are dummy variables, and therefore one needs to be removed or 148 149 suppressed in order for the analysis to be conducted. The procedure would break down if all the dummies were used together. This is due to the fact that once one knows the values for the first twelve variables, the value of the thirteenth is known with certainty. Justice Harlan was chosen because he and Justice Black participated in the fewest number of decisions (14), and Justice Black is more closely associated with this area of law due to his opinion in the Everson case. Furthermore, it should be noted that analysis was also done with Harlan included and Black removed with no significant difference in the results. 3Having positive and negative variable coefficients causes no ill effects on the model as a whole. They simply influence the model in different directions. 4Justice O’Connor was not, of course, part of Spaeth’s 1979 research. Her voting record on the Court appears to be clearly "conservative" and therefore she was placed with the other justices whom Spaeth labeled as having such a value system. 5The ten cases are: National Labor Relations Board L. Catholic Bishop of Chicago (1979), Alamo Foundation 1:. Secretary of Labor (1985), St. Martin Evangelical Lutheran Church v. South Dakota (1981), McDaniel v. Paty (1978), Bob JOnes University v. U.S. (1983), Goldsboro Christian Schools v. U.S. (1983), Harris v. McRae (1980), Welsh y; U.S. (1970), Gillette v. U.S. (1971), and Negre v. Larson (1971). 6The two cases are: Treen v. Karen B. (1982), and Norwood v. Harrison (1973). 7See note 1. 8The data for this research, as was stated above, are not from a random sample. Instead, this research attempts to examine the entire population of relevant data, and there- fore the violation of this assumption should not be a con- cern. References Abraham, Henry. 1980. The Judicial Process. New York: Oxford University Press. Abraham, Henry. 1987. The Judiciary: The Supreme Court in the Governmental Process, 7th ed. Boston: Allyn and Bacon. Aldrich, John, and Forrest Nelson. 1984. Linear Probability, Logithand Probit Models. Beverly Hills: Sage. Aldrich, John, Gary Miller, Charles Ostrom, Jr., and David Rohde. 1986. American Government: People, Institutions, and Policies. Boston: Houghton Mifflin Company. Alexander, Christopher. 1963. Notes on the Synthesis of Form. Cambridge: Harvard University Press. Allison, Graham T. 1971. Essence of Decision. Boston: Little Brown and Company. Armstrong, Virginia, and Charles Johnson. 1982. Certiorari Decisions by the Warren and Burger Courts: Is Cue Theory Time Bound? Polit , 15:143-150. Ashby, W. Ross. 1956. An Introduction to Cybernetics. New York: John Wiley & Sons. Baum, Lawrence. 1985. The Supreme Court. 2nd ed. Washington D.C.: Congressional Quarterly Press. Becker, Theordore. 1963. Inquiry into a School of Thought in the Judicial Behavior Movement. Midwest Journal of Politi- cal Science, 7:253-266. Beer, Stafford. 1964. Cybernetics and Management. New York: John Wiley & Sons. Bickel, Alexander. 1973. The Caseload of the Supreme Court and What, if Anything, tg_po About 12- Buchanan, Sidney. 1978. Governmental Aid to Sectarian Schools: A Study in Corrosive Precedents. Houston Law Review, 15:835. Caldeira, Gregory, and John Wright. 1988. Amici Curiae Before the Supreme Court: Who Participates, When, and How Much? Paper delivered at the annual meeting of the Midwest Political Science Association. Cannon, Bradley, and Michael Giles. 1972. Recurring Litigants: Federal Agencies Before the Supreme Court. Western Political Science Quarterly, 25:183-191. 150 151 Caplan, Lincoln. 1987. The Tenth Justice: the Solicitor Gen- eral and the Rule of Law. New York: Knopf. Carrington, D. 1974. U.S. Appeals in Civil Cases: A Field and Statistical Study. Houston Law Review, 11:1101-1129. Casper, Gerhard, and Richard Posner. 1976. The Workload of the Supreme Court. Chicago: American Bar Foundation. Casper, Jonathan. 1977. The Politics of Civil Liberties. New York: Harper and Row. Chopper, Jesse. The Religion Clauses of the First Amendment: Reconciling the Conflict. University of Pittsburg Law Review, 41:680. Chopper, Jesse. 1986. In Encyclopedia of the American Con- stitution, ed. Leonard Levy. New York: Macmillian. Cohen, William, and John Kaplan. 1982. Constitutional Law. 2nd ed. Mineola, NY: The Foundation Press. Cord, Robert. 1982. Separation of Church and State. New York: Lambeth Press. De Tocqueville, Alexis. 1945. Democracysin America. rev.ed. New York: Alfred A. Knopf. Everson v. Board of Education. 1947. 330 U.S. 1. Flango, Victor, and Craig Ducat. 1977. Toward an Integration of Public Law and Judicial Behavior. Journal of Politics, 39:41-72. Frohlich, Norman, and Joe A. Oppenheimer. 1978. Modern Political Economy. Englewood Cliffs: Prentice-Hall. Gaffney, Edward. 1980. Political Divisiveness Along Religious Lines: the Entanglement of the Court in Sloppy History and Bad Public Policy. St. Louis University Law Journal, 24:205-236. Galanter, Mark. 1974. Why the "Haves" Come Out Ahead: Speculation on the Limits of Legal Change. Law and Society Review, 9:95-160. Gemmer, David. 1982. The Religion and Speech Clauses: A New Imbalance. Stetson Law Review, 12:155-181. George, A.L. 1980. Presidential Decision-making in Foreign Policy: The Effective Use of Information and Advice. Boulder, Colorado: Westview. 152 Gibson, James. 1978. Judges’ Role Orientations, Attutudes, and Decisions: An Interactive Model. American Political Science Review, 72:911-924. Gibson, James. 1983. From Simplicity to Complexity: The Development of Theory in the Study of Judicial Behavior. Political Behavior, 5:7-49. Goldman, Sheldon, and Thomas Jahnige. 1976. The Federal Courts as s Politicsl System. 2nd ed. New York: Harper & Row. Goldman, Sheldon. 1979. The Effect of Past Judicial Behavior on Subsequent Decision-Making. Jurimetrics Journal, 19:208-217. Goldman, Sheldon, and Austin Sarat, eds. 1989. American Court Systems. New York: Longman. Grossman, Joel, and Richard Wells. 1988. Constitutional Law and Judicial Policy Making. 3rd ed. New York: Longman. Gryski, Gerard, Eleanor Main, and William Dixon. 1986. Models of State High Court Decision Making in Sex Discrimination Cases. Journal of Politics. 48:143-153. Hayakawa, Takeo. 1962. Legal Science and Judicial Behavior with Particular Reference to Civil Liberties in the Japanese Supreme Court. Kobe University Law Review, 2:1- 27. Henschen, Beth. 1983. Statutory Interpretation of the Supreme Court--Congressional Response. American Politics Quarterly, 11:441-458. Howard, J. Woodford. 1982. Query: Are Heavy Caseloads Chang- ing the Nature of Appellate Justice. Judicature, 66:57—59. Howe, Mark DeWolfe. 1965. The Garden snd the Wilderness: Religion and Government in American Constitutional His- tory. Chicago: University of Chicago Press. Hutchinson, Dennis. 1979. Unanimity and Desegregation: Decision-Making in the Supreme Court, 1948-1958. Georgetown Law Journal, 68:1-96. Kauper, Paul. 1964. Religion and the Constitution. Baton Rouge: L.S.U. Press. Kort, Fred. 1957. Predicting Supreme Court Decisions Mathe- matically: A Quantitative Analysis of the Right to Counsel Cases. American Political Science Review, 51:1-12. 153 Kort, Fred. 1963. Content Analysis and Judicial Opinions and Rules of Law. In Judicial Decision Making, ed. Glendon Schubert. New York: The Free Press of Glencoe. Kort, Fred. 1964a. Comment on "The Untroubled World of Jurimetrics". Journal of Politics, 26:923-926. Kort, Fred. 1964b. Mathematical Models of the Decision Func- tion. In Judicial Behavior: A Reader in Theory and Research, ed. Glendon Schubert. Chicago: Rand McNally. Kort, Fred. 1966. Quantitative Analysis of Fact-Patterns in Cases and their Impact on Judicial Decisions. Harvard Law Review, 79:1595-1603. Kort, Fred. 1968. A Nonlinear Model for the Analysis of Judi- cial Decisions. American Political Science Review, 62:546- 555. Kort, Fred. 1973. Regression Analysis and Discriminant Analy- sis: An Application of R.A. Fisher’s Theorem to Data in Political Science. American Political Science Review, 67:555-559. Kurland, Philip. 1961. Of Church and State and the Supreme Court. University of Chicago Law Review, 29:1-96. Larson, D.W. 1985. Origins of Containment: A Psychological Explanation. Princeton, NJ: Princeton University Press. Lau, Richard, and David Sears. 1986. Political Cognition. Hillsdale, NJ: Erlbaum. Lawlor, Reed. 1964. Stare Decisis and Electronic Computers. In Judicial Behavior: A Reader in Theory and Research, ed. Glendon Schubert. Chicago: Rand McNally. Lawlor, Reed. 1967. Personal Stare Decisis. Southern California Law Review, 41:73-118. Levy, Leonard. 1986. The Establishment Clause: Religion and the First Amendment. New York: Macmillian Publishing Co. Lewis, Anthony. 1964. Gideon’s Trumpet. New York: Random House. Lindblom, Charles. 1959. The Science of Muddling Through. Public Administration Review, 19:79-88. Luce, R. Duncan, and Howard Raiffa. 1957. Games and Deci- sions. New York: John Wiley & Sons. 154 Markus, H., and R.B. Zajonc. 1985. The Cognitive Perspective in Social Psychology. In The Handbook of Social Psychol- ggy. 3rd ed. eds. G. Lidzey and E. Aronson. New York: Ran- dom House. Mawdsley, Ralph. 1986. Prohibition of Student Religious Activities on Public School Premises: Unreasonable Content-Based Restriction. Cooley Law Review, 4:117-157. McCarthy, Martha. 1983. A Delicate Balance: Church, State and the Schools. Bloomington, Indiana: Phi Delta Kappen Educa- tional Foundation. Mendelson, Wallace. 1963. The Neo-Behavior Approach to the Judicial Process: A Critique. American Political Science Review, 57:593-603. Mendelson, Wallace. 1964a. The Untroubled World of Jurimetrics. Journal of Politics, 26:914-922. Mendelson, Wallace. 1964b. Response by Professor Mendelson. Journal of Politics, 26:927-928. Mendelson, Wallace. 1966. An Open Letter to Professor Spaeth and His Jurimetrical Colleagues. Journal of Politics, 28:429-432. Murphy, Walter. 1964. Elements of Judicial Strategy. Chicago: University of Chicago Press. Nagel, Stuart. 1960. Using Simple Calculations to Predict Judicial Decisions. American Behavioral Scientist, 4: 24- 28. Nowak, John, Ronald Rotunda, and J. Nelson Young. 1978. Constitutional Law. St. Paul, Minn.:West Publishing Co. O’Brien, David. 1986. Storm Center. New York: W.W. Norton & Company. O’Connor, Karen. 1983. The Amicus Curiae Role of the U.S. Solicitor General in Supreme Court Litigation. Judicature, 66:256-264. O’Connor, Karen, and Lee Epstein. 1983. States Rights or Criminal Rights: An Analysis of State Performance in U.S Supreme Court Litigation. Paper presented at the annual meeting of the Northeastern Political Science Association. Ostrom, Charles Jr., and Brian Job. 1986. The President and the Political Use of Force. American Political Science Review, 80:542-566. 155 Ostrom, Charles Jr., and Dennis Simon. The President’s Pub- lic. American Journal of Political Science, (forthcoming). Pfeffer, Leo. 1967. Church. State. and Freedom. Boston: Beacon Press. Pfeffer, Leo. 1979. The Current State of the Law in the U.S and the Separationist Agenda. The Annals of the American Academy of Political and Social Science. 1979:1-9. Pfeffer, Leo. 1980. Freedom and/or Separation: The Constitu- tional Dilemma of the First Amendment . Minnesota Law Review, 64: 561-584. Pfeffer, Leo. 1984. Religion. State and the Burger Court. Buffalo, N.Y.: Prometheus Books. Pindyck, Robert, and Daniel Rubinfeld. 1976. Econometric Models and Economic Forecasts. New York: McGraw-Hill. Pritchett, C. Herman. 1941. Divisions of Opinion Among Jus- tices of the U.S. Supreme Court, 1939-1941. American Political Science Review. 35:890-897. Pritchett, C. Herman. 1948. The Roosevelt Court: A Study in Judicial Politics and Values. 1937-1947. New York: Macmil- lan. Puro, Steven. 1981. The United States as Amicus Curiae. In Courts. Law, and the Judicial Process, ed. S. Sidney Ulmer. New York: Free Press. Ripple, Kenneth. 1980a. The Entanglement Test of the Religion Clauses--A Ten Year Assessment. UCLA Law Review, 27:1195- 1239. Ripple, Kenneth. 1980b. The Supreme Court’s Workload: Some Thoughts for the Practitioner. American Bar Association Journal, 66:174-176. Roche, John. 1958. Political Science and Science Fiction. Americal Political Science Review, 52:1026-1029. Rohde, David. 1972. Policy Goals and Opinion Coalitions in the Supreme Court. Midwest Journal of Political Science, 16:208-224. Rohde, David, and Harold Spaeth. 1976. Supreme Court Decision Making. San Francisco: W.H. Freeman and Co. Samuelson, Paul. 1970. Economics. New York: McGraw-Hill Book Company. 156 Schubert, Glendon. 1958. The Study of Judicial Decision- Making as an Aspect of Political Behavior. American Political Science Review, 52:1007-1025. Schubert, Glendon. 1959. Quantitative Analysis of Judicial Behavior. Glencoe, IL: The Free Press. Schubert, Glendon. 1961. A Psychometric Model of the Supreme Court. American Behavioral Scientist, 5:14-18. Schubert, Glendon. 1962. The 1960-61 Term of the Supreme Court: A Psychometric Analysis. American Political Science Review, 56:90-101. Schubert, Glendon. 1963. Behavioral Research in Public Law. American Polical Science Review, 57:433-445. Schubert, Glendon. 1964. Judicial Behavior: A Reader in Theory and Research. Chicago: Rand McNally. Schubert, Glendon. 1965. The Judicial Mind: Attitudes and Ideologies of Supreme Court Justices. 1946-1963. Evanston, IL: Northwestern University Press. Schubert, Glendon. 1967. Ideologies and Attitudes, Academic and Judicial. Journal of Politics, 26: 3-41. Schubert, Glendon. 1974. The Judicial Mind Revisited. New York: Oxford University Press. Scigliano, Robert. 1971. The Supreme Court and the Presi- dency. New York: Free Press. Segal, Jeffrey. 1984. Predicting Supreme Court Cases Probabilistically: The Search and Seizure Cases,l962-1981. American Political Science Review, 78:891-900. Segal, Jeffrey. 1985. Measuring Change on the Supreme Court: Examining Alternative Models. American Journal of Politi- cal Science, 29:461-478. Segal, Jeffrey. 1986. Supreme Court Justices as Human Deci— sion Makers: An Individual-Level Analysis of the Search and Seizure Cases. Journal of Politics, 48:938. Segal, Jeffrey. 1988. Amicus Curiae Briefs by the Solicitor General during the Warren and Burger Courts: A Research Note. The Western Political Quarterly, 41:135-144. Shapiro, Martin. 1972. Toward a Theory of Stare Decisis. Journal of Legal Studies, 1:125-134. Simon, Herbert. 1957. Models of Man. New York: John Wiley & Sons. 157 Simon, Herbert. 1959. Theories of Decision Making in Econom- ics. American Economic Review, 49:253-283. Simon, Herbert. 1979. Models of Thought. New Haven: Yale University Press. Simon, Herbert. 1981. Sciences of the Artificial. 2nd ed. Cambridge: MIT Press. Simon, Herbert. 1985. Human Nature in Politics: The Dialogue of Psychology with Political Science. American Political Science Review, 79:293-304. Songer, Donald. 1979. Concern for Policy Outputs as a Cue for Supreme Court Decisions on Certiorari. Journal of Politics, 41:1185-1194. Spaeth, Harold. 1961. An Approach to the Study of Attitudinal Differences as an Aspect of Judicial Behavior. Midwest Journal of Political Science, 5:165-180. Spaeth, Harold. 1962. Judicial Power as a Variable Motivating Supreme Court Behavior. Midwest Journal of Political Science, 6:54-82. Spaeth, Harold. 1963a. Warren Court Attitudes toward Busi- ness: The ’B’ scale. In Judicial Decision-Making, ed. Glendon Schubert. New York: Free Press. Spaeth, Harold. 1963b. An Analysis of Judicial Attitudes in the Labor Relations Decisions of the Warren Court. Journal of Politics, 25:290-311. Spaeth, Harold. 1964. The Judicial Restraint of Mr. Justice Frankfurter: Myth or Reality? Midwest Journal of Political Science, 8:22-38. Spaeth, Harold. 1965. Jurimetrics and Professor Mendelson: A Troubled Relationship. Journal of Politics, 27:875-880. Spaeth, Harold, and David Peterson. 1971. The Analysis of Interpretation of Dimensionality: The Case of Civil Liberty Decision Making. Midwest Journal of Political Science, 15: 415-441. Spaeth, Harold. 1979. Supreme Court Policy Making. San Fran- cisco: W.H. Freeman and Co. Steinbruner, John D. 1974. The Cybernetic Theory of Decision. Princeton, NJ: Princeton University Press. Stumpf, Harry. 1988. American Judicial Politics. New York: Harcourt, Brace, Jovanovich. 158 Tager, Evan. 1984. The Supreme Court, Effect Inquiry, and Aid to Parochial Education. Stanford Law Review, 37:219-251. Tanenhaus, Joseph. 1960. Supreme Court Attitudes Toward Administrative Agencies. Journal of Politics, 22:502-524. Tanenhaus, Joseph . 1961. Supreme Court Attitudes toward Fed- eral Administrative Agencies 1947-1956--An Application of Social Science Methods to the Study of the Judicial Process. Vanderbilt Law Review, 14: 482-502. Tanenhaus, Joseph, M. Schick, M. Muraskin, and D.Rosen. 1963. The Supreme Court’s Certiorari Jurisdiction and Cue Theory. In Judicial Decision Making, ed. Glendon Schubert. Glencoe, IL.: Free Press. Tanenhaus, Joseph. 1966. The Cumlative Scaling of Judicial Decisions. Harvard Law Review, 79:1583-94. Tate, Neal. 1983. The Methodology of Judicial Behavior Research: A Review and Critique. Political Behavior, 5:51- 82. Teger, Stuart. 1977. Presidential Strategy for the Appoint- ment of Supreme Court Justices. Public Choice, 31:1-22. Teger, Stuart, and Douglas Kosinski. 1980. The Cue Theory of Supreme Court Certiorari Jurisdiction: A Reconsideration. Journal of Politics, 42: 834-846. Tribe, Laurence. 1978. American Constitutional Law. Mineola, NY: The Foundation Press. Ulmer, S. Sidney. 1960. The Analysis of Behavior Patterns in the United States Supreme Court. Journal of Politics, 22:629-653. Ulmer, S. Sidney. 1961a. A Note on Attitudinal Consistency in the United States Supreme Court. Indian Journal of Politi- cal Science, 22:195-204. Ulmer, S. Sidney. 1961b. On the Scaling of Judicial Decisions (Communication to the Editor). American Political Science Review, 55:599-600. Ulmer, S. Sidney. 1962. Supreme Court Behavior in Racial Exclusion Cases: 1935-1960. American Political Science Review, 56:325-330. Ulmer, S. Sidney. 1964. Discriminant Analysis and an Error Criterion. In Judicial Behavior: A Reader in Theory and Research, ed. Glendon Schubert. Chicago: Rand McNally. 159 Ulmer, S. Sidney. 1969. The Dimensionality of Judicial Voting Behavior. Midwest Journal of Political Science, 13:471- 483. Ulmer, S. Sidney. 1971. Earl Warren and the Brown Decision. Journal of Politics, 33: 692-696. Ulmer, S. Sidney, William Hintze, and Louise Kirklosky. 1972. The Decision to Grant or Deny Certiorari: Further Con- sideration of Cue Theory. Law and Society Review, 6:637- 643. Ulmer, S. Sidney. 1981. The Supreme Court and ’Jury Selec- tion’ Facts. In Courts, Lastand Judicial Processes, ed. 8. Sidney Ulmer. New York: The Free Press. Ulmer, S. Sidney. 1984. The Supreme Court’s Certiorari Deci- sions: Conflict as a Predictive Variable. American Politi- cal Science Review, 78:902-911. Ulmer, S. Sidney and David Willison. 1985. The Solicitor Gen- eral of the U.S. as Amicus Curiae in the U.S. Supreme Court, 1969-1983 Terms. Paper presented at the annual meeting of the American Political Science Association. Ulmer, S. Sidney. 1986. Sppreme Court Policymaking and Con- stitutional Law. New York: McGraw-Hill Book Co. U.S. Department of Justice. 1985. Annual Report of the Attorney General of the United States. 1984. Washington: U.S. Government Printing Office. U.S. v. Butler. 1936. 297 U.S. l. Valente, William. 1969. Aid to Church-Related Education--New Directions without Dogma. Virginia Law Review, 55:579. Van Alstyne, William. 1984. Trends in the Supreme Court: Mr. Jefferson’s Crumbling Wall--A Comment on Lynch V. Don- nelly. Duke Law Journal, 1984:770-787. Van Patten, Jonathan. 1983. In the End Is the Begining: An Inquiry Into the Meaning of the Religion Clauses. St. Louis University Law Journal, 27: 1-93. Wasby, Stephen. 1984. The Supreme Court in the Federal Judi- cial System. New York: Holt, Rinehart, & Winston. Werdegar, Kathryn Mickle. 1967. The Solicitor General and Administrative Due Process. George Washington Law Review, 36:481-514. 160 Wiener, N. 1948. Cybernetics. New York: John Wiley & Sons. Woodward, Bob, and Scott Armstrong. 1979. The Brethren. New York: Simon and Schuster.